E-Procurement Adoption by Suppliers: A Research Proposal

Bráulio Alturas

ABSTRACT This paper presents a current research with the aim to identify the enablers, barriers and possible solutions to eprocurement adoption by suppliers in Portugal. The literature relating to e-procurement implementation and operation is reviewed, with focus on the barriers and enablers already identified in the literature. A research methodology is proposed to study the problem, and this work will contribute to better address the issues faced by suppliers on e-procurement implementations.

call for papers IADIS MULTI CONFERENCE ON COMPUTER SCIENCE AND INFORMATION SYSTEMS 19-21 June Algarve, Portugal Proceedings of e-COMMERCE 2009 Edited by: Sandeep Krishnamurthy international association for development of the information society IADIS INTERNATIONAL CONFERENCE E-COMMERCE 2009 part of the IADIS MULTI CONFERENCE ON COMPUTER SCIENCE AND INFORMATION SYSTEMS 2009 ii PROCEEDINGS OF THE IADIS INTERNATIONAL CONFERENCE E-COMMERCE 2009 part of the IADIS MULTI CONFERENCE ON COMPUTER SCIENCE AND INFORMATION SYSTEMS 2009 Algarve, Portugal JUNE 19 - 21, 2009 Organised by IADIS International Association for Development of the Information Society iii Copyright 2009 IADIS Press All rights reserved This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks. Permission for use must always be obtained from IADIS Press. Please contact secretariat@iadis.org e-Commerce Volume Editor: Sandeep Krishnamurthy Computer Science and Information Systems Series Editors: Piet Kommers, Pedro Isaías and Nian-Shing Chen Associate Editors: Luís Rodrigues and Patrícia Barbosa ISBN: 978-972-8924-89-8 SUPPORTED BY iv TABLE OF CONTENTS FOREWORD ix PROGRAM COMMITTEE xi KEYNOTE LECTURE xv FULL PAPERS BARRIERS OF ONLINE SHOPPING IN DEVELOPING COUNTRIES: CASE STUDY IRAN Omid Bigdeli, Sanam Afaghzadeh, Amir Albadvi and Neda Abdolvand FACTORS AFFECTING CONSUMERS ADOPTION OF ECOMMERCE IN SAUDI ARABIA FROM A CONSUMERS’ PERSPECTIVE 3 11 Fahad Aleid, Simon Rogerson and Ben Fairweather DOES FIT B2B E-COMMERCE FOR AGRIBUSINESS? A FIRST APPROACH FOR TRUST EVALUATION ISSUES IN SPANISH AGRIFOOD SECTOR 19 Mª Cristina Fernández, Isabel de Felipe, Julián Briz STRATEGIC ALIGNMENT AS A WAY OF ADDRESSING THE BARRIERS TO E-BUSINESS ADOPTION 27 Eduardo Escofet, María José Rodríguez, José Luis Garrido and Lawrence Chung ECUSTOMS CASE STUDY: MECHANISMS BEHIND CO-OPERATION PLANNING 35 María Laura Ponisio, Pascal van Eck and Lourens Riemens FINDING TREND LEADERS FROM MARKETING TRANSACTION DATA 45 Masakazu Takahashi, Kazuhiko Tsuda and Takao Terano ADDING VALUE TO ENTERPRISEWIDE SYSTEM INTEGRATION: A NEW THEORETICAL FRAMEWORK FOR ASSESSING TECHNOLOGY ADOPTION OUTCOMES 53 Linda Wilkins, Paula M.C. Swatman and Duncan Holt IEBS: A MODEL FOR INTELLIGENT E-BIDDING 61 Pedro Brandão Neto, Sofiane Labidi, Rafael de Souza Cunha and Rafael Soares Cruz AN APROACH FOR SEMANTIC-BASED EC MIDDLEWARE Ejub Kajan and Leonid Stoimenov v 69 SMART SHOPPING SPACES: CONNECTING MERCHANTS AND CONSUMERS BY INNOVATIVE ONLINE MARKETPLACES 77 Peter Leitner and Thomas Grechenig FROM A RETAIL-BASED E-MICROPAYMENT PROGRAM TO A MULTIPURPOSE PROGRAM: WHICH CRITICAL FACTORS ARE NEEDED FOR SUCH A TRANSFORMATION? 89 Wee Kheng-Tan and Yu-Jie Tan SITE PERSONALIZATION PROCESS BASED ON NAVIGATIONAL BEHAVIOR AND FUZZY ONTOLOGY 97 Juliano Z. Blanco, Antonio Francisco do Prado A CONTENT ANALYSIS OF WEB-SITE QUALITY OF ON-LINE AUCTION SELLERS 105 Fen-Hui Lin and Chiu-Chu Hwang GUIDELINES TO THE DEVELOPMENT OF AN E-COMMERCE PLATFORM FOR CUSTOMIZED GARMENTS 113 Liliana Ribeiro, Paulo Duarte and Rui Miguel TIME AND SPACE CONTEXTUAL INFORMATION IMPROVES CLICK QUALITY ESTIMATION 123 Mehmed Kantardzic, Brent Wenerstrom, Chamila Walgampaya, Oleksandr Lozitskiy, Sean Higgins, Darren King THE BRAZILIAN MOBILE DIGITAL CONTENT MARKET: AN OVERVIEW 131 Marcelo Cortimiglia, Filippo Renga and Andrea Rangone BUYER’S MINDSET ABOUT ONLINE PURCHASE AND AUCTION, AND ITS EFFECT ON PAYMENT CHOICE 139 Wee Kheng Tan and Yung Lun Chung IMPROVING DIVERSITY AND RELEVANCY OF E-COMMERCE RECOMMENDER SYSTEMS THROUGH NLP TECHNIQUES 147 Andriy Shepitsen and Noriko Tomuro FACTORS AFFECTING ONLINE APPLICATION OF INSURANCE PRODUCTS AND ITS IMPLICATIONS Wee Kheng-Tan and Yu-Jie Tan 155 THE CONFLICTING ROLE OF ONLINE SWITCHING COSTS: THE MAIN AND INTERACTION EFFECTS ON CUSTOMER RETENTION 163 Ana Isabel Torres and Francisco Vitorino Martins RECOGNITION OF EMOTIONS IN E-COMMERCE-SETTINGS Susanne Robra-Bissantz vi 171 SHORT PAPERS E-PROCUREMENT ADOPTION BY SUPPLIERS: A RESEARCH PROPOSAL 183 Paulo Andrade and Bráulio Alturas MEASURING VIRTUAL KNOWLEDGE MANAGEMENT IMPACT IN FIRM’S PERFORMANCE 188 Flávio Gomes Borges Tiago, Maria Teresa Borges Tiago and João Pedro Almeida Couto MODERN ARCHITECTURAL REASONING FOR COMPLEX WEB COMMERCE APPLICATIONS 193 Thomas Lehrner, Birgit Pohn, Markus Schranz A NEGOTIATION MECHANISM USING ARGUMENT-BASED METHODS AND PERIPHERAL ISSUES IN MULTI-AGENT SYSTEMS WITH INCOMPLETE INFORMATION Nathalie Hrycej and Azita Darooei, Mohammad-Reza 199 Khayyambashi SELF-PRODUCT CONGRUENCE: IMAGE-PERCEPTIONS OF POSTMODERN OUTDOOR-APPAREL CONSUMERS IN E-COMMUNITIES 205 Jan Breitsohl and Marwan Khammash EXTERNALIZATION OF VIRTUAL PROTOTYPES AS AN E-COMMERCE SERVICE IN THE FASHION INDUSTRY 210 Carolin Löffler BUILDING FINANCIAL CAPABILITY VIA THE INTERNET 215 Tanai Khiaonarong A CUSTOMER FOCUSED E-COMMERCE APPROACH USING PURCHASING WEB PATTERNS 219 Markus Weinmann and Yvonne Gaedke ASSESSING THE CONTRIBUTION OF SCM ON E-BUSINESS PERFORMANCE 224 Maria Teresa Borges Tiago, João Pedro Couto and Flávio Tiago INTERORGANIZATIONAL BUSINESS PROCESSES MODELING: CASE OF AN E-PROCUREMENT SYSTEM Khoutir Bouchbout and Zaia Alimazighi 229 POSTERS WEB AND INFORMATION TECHNOLOGY AS CRITICAL THEMES IN THE CONSUMER BEHAVIOR BASED RESEARCH: A STUDY WITH THEMATIC NETWORKS María Isabel Viedma-del-Jesus, Antonio Gabriel López-Herrera, Juan Sánchez-Fernández and Francisco Muñoz-Leiva vii 237 THE ANTECEDENTS OF USEFULNESS IN EXPERIENCED USERS OF WEB-BASED LEARNING MANAGEMENT SYSTEMS 239 Francisco Muñoz Leiva, Juan Sánchez-Fernández, María Isabel Viedma-del-Jesús, Antonio Gabriel López-Herrera 242 APPLICATION OF TERNARY AHP Sylvia Encheva DOCTORAL CONSORTIUM FLOW AND ONLINE CONSUMER BEHAVIOUR: AN EMPIRICAL ANALYSIS OF E-LEARNING EXPERIENCES 247 Irene Esteban-Millat, Inma Rodríguez-Ardura and Antoni Meseguer THE IMPACT OF CULTURAL ADAPTATION ON THE EFFECTIVENESS OF E-COMMERCE WEBSITES Femke Vyncke, Malaika Brengman and Olga De Troyer AUTHOR INDEX viii 251 FOREWORD These proceedings contain the papers of the IADIS International Conference on eCommerce 2009, which was organised by the International Association for Development of the Information Society in Algarve, Portugal, 19 – 21 June, 2009. This conference is part of the Multi Conference on Computer Science and Information Systems 2009, 17 - 23 June 2009, which had a total of 1131 submissions. IADIS e-Commerce 2009 conference is a major international event for researchers, academics, industry specialists, practitioners & students interested in the advances in, and applications of, eCommerce. The participants will have an opportunity to present and observe the latest research results, and ideas in these areas. This conference aims to cover both technological as well as non-technological issues related to this new business paradigm. The Conference invites proposals from the introductory through advanced level on all topics related to e-Commerce. Proposals which address the theory, research and applications as well as describe innovative projects are encouraged. The following five main areas have been the object of paper and poster submissions within specific topics: - e-Commerce Technology; - Global e-Commerce; - Online Management; - Online Business Models; - Regulatory/Policy Issues. The IADIS e-Commerce 2009 received 92 submissions from more than 26 countries. Each submission has been anonymously reviewed by an average of four independent reviewers, to ensure that accepted submissions were of a high standard. Consequently only 21 full papers were approved which means an acceptance rate below 23 %. A few more papers were accepted as short papers, reflection papers, doctoral consortium and posters. An extended version of the best papers will be published in the IADIS International Journal on WWW/Internet (16457641), IADIS International Journal on Computer Science and Information Systems (ISSN: 1646-3692), in a special issue of the DMIJ - Direct Marketing: An International Journal (ISSN: 1750-5933) and also in other selected journals, including journals from Inderscience. Besides the presentation of full papers, short papers, reflection papers, doctoral consortium and posters, the conference also included one keynote presentation from an internationally distinguished researcher. We would therefore like to express our gratitude to Dr. Shintaro Okazaki, Autonomous University of Madrid, Spain. As we all know, organising a conference requires the effort of many individuals. We would like to thank all members of the Program Committee, for their hard work in reviewing and selecting the papers that appear in the proceedings. ix This volume has taken shape as a result of the contributions from a number of individuals. We are grateful to all authors who have submitted their papers to enrich the conference proceedings. We wish to thank all members of the organizing committee, delegates, invitees and guests whose contribution and involvement are crucial for the success of the conference. Last but not the least, we hope that everybody will have a good time in Algarve, and we invite all participants for the next year edition of the IADIS International Conference on eCommerce 2010, that will be held in Freiburg, Germany. Sandeep Krishnamurthy, Program Chair University of Washington, USA e-Commerce 2009 Conference Program Chair Piet Kommers, University of Twente, The Netherlands Pedro Isaías, Universidade Aberta (Portuguese Open University), Portugal Nian-Shing Chen, National Sun Yat-sen University, Taiwan MCCSIS 2009 General Conference Co-Chairs Algarve, Portugal June 2009 x PROGRAM COMMITTEE E-COMMERCE CONFERENCE PROGRAM CHAIR Sandeep Krishnamurthy, University of Washington, USA MCCSIS GENERAL CONFERENCE CO-CHAIRS Piet Kommers, University of Twente, The Netherlands Pedro Isaías, Universidade Aberta (Portuguese Open University), Portugal Nian-Shing Chen, National Sun Yat-sen University, Taiwan E-COMMERCE CONFERENCE COMMITTEE MEMBERS Adam Vrechopoulos, Athens University of Economics and Business, Greece Adina Magda Florea, Politehnica University of Bucharest, Romania Agnes Koschmider, University Karlsruhe, Germany Alex Norta, University of Helsinki, Finland Andrea Carignani, University of Milan, Italy Andreas Ekelhart, Security Research, Austria Andrew Byde, HP Labs Bristol, UK Andy Phippen, University of Plymouth, UK Antoni Meseguer-Artola, Universitat Oberta de Catalunya, Spain Antonio López Herrera, University of Granada, Spain Aristogiannis Garmpis, Technological Educational Institution of Messolonghi, Greece Arthur Csetenyi, Budapest Corvinus University, Hungary Blanca Hernández, Universidad de Zaragoza, Spain Borislav Josanov, Novi Sad Business School, Serbia Božo Matic, Faculty of Economics and Business, Croatia Carla Ruiz Mafe, University of Valencia, Spain Christian Schloegl, University of Graz, Austria Christos Xenakis, University of Piraeus, Greece Costas Lambrinoudakis, University of the Aegean, Greece Costin Badica, University of Craiova, Romania Damminda Alahakoon, Monash University, Australia Daniel Perez-Gonzalez, University of Cantabria, Spain Dennis Tachiki, Tamagawa University, Japan Dimitris Geneiatakis, University of the Aegean, Greece Dimitris Kanellopoulos, University of Patras, Greece Dimitris Rigas, University of Bradford, United Kingdom Eduard Cristóbal, Universitat de Lleida, Spain xi Eduardo Peis, University of Granada, Spain Ejub Kajan, High School of Applied Sudies, Serbia Eliana Rocío Rocha, Universidad de Cantabria, Spain Emmanouel Varvarigos, University of Patras, Greece Emulija Vuksanovic, Faculty For Economics, Kragujevac, Serbia Enrique Bigné, Universidad de Valencia, Spain Enrique Herrera Viedma, University of Granada, Spain Euripidis Loukis, University of the Aegean, Greece Eva Rimbau Gilabert, Oberta University of Catalonia, Spain Fotis Lazarinis, Technological Educational Institute of Mesolonghi, Greece Francisco J. Martínez-López, University of Granada, Spain Franz Lehner, Universität Passau, Germany George Dafoulas, Middlesex University, United Kingdom George Kambourakis, University of the Aegean, Greece Gerard Ryan, Universitat Rovira i Virgili, Spain Hana Horak, Faculty of Economics and Business, Croatia Heiko Pfeffer, Fraunhofer Institut FOKUS, Germany Hiroaki Fukuda, Keio University, Japan Inma Rodríguez-Ardura, Universitat Oberta de Catalunya, Spain Isabel de Felipe, Universidad Politécnica de Madrid, Spain Isabella Mader, IMAC Information & Management Consulting, Germany Ivan Strugar, Universitiy of Zagreb, Croatia Jacob Carsten, Fraunhofer Institut FOKUS, Germany Jemal H. Abawajy, Deakin University, Australia Jens Fromm, Fraunhofer FOKUS, Germany Jeroen Doumen, Irdeto, The Netherlands Jose Manuel Morales del Castillo, Universidad de Granada, Spain Josef Herget, Danube University Krems, Austria Joseph Heili, Chambery School of Business, France Julián Briz, Universidad Politécnica de Madrid, Spain Jun Suzuki, University of Massachusetts, USA Jurica Pavicic, University of Zagreb, Croatia Kamel Rouibah, College of Business Administration, Kuwait Krassie Petrova, Auckland University of Technology, New Zealand Luisa Andreu, Universitat de València, Spain Mar Pàmies, Universitat Rovira i Virgili, Spain Marc Esteva, IIIA-CSIC, Spain Marco Furini, University of Piemonte Orientale, Italy Marco Mevius, FZI, Research Center for Information Technologies, Germany Margarita Alonso, Universidad de Cantabria, Spain Maria Papadaki, University of Plymouth, UK Mario Spremic, University of Zagreb, Croatia Markus Schranz, Vienna University of Technology, Austria Martin Smits, Tilburg University, The Netherlands Mary Tate, Victoria University of Wellington, New Zealand Masitah Ghazali, Universiti Putra Malaysia, Malaysia Masrah Azrifah Azmi Murad, Universiti Putra Malaysia, Malaysia xii Matjaz Gams, Jozef Stefan Institute, Slovenia Michael Merz, Ponton Consulting, Germany Michelangelo Ceci, University of Bari, Italy Minseok Song, Eindhoven University of Technology, The Netherlands Nahed Azab, Middlesex University, United Kingdom Nineta Polemi, University of Piraeus, Greece Noor Akma Mohd Salleh, University Malaya, Malaysia Nordin bin Zakaria, Universiti Teknologi Petronas, Malaysia Oshadi Alahakoon, Monash University, Australia Ota Novotny, University of Economics, Czech Republic Pedro Solana González, Universidad de Cantabria, Spain Pedro Soto-Acosta, Universidad de Múrcia, Spain Pere Tumbas, Faculty of Economics University of Novi Sad, Serbia Peter Weiß, University of Karlsruhe, Germany Petra Hoepner, Fraunhofer Institut FOKUS, Germany Rainer Schmidt, Aalen University, Germany Rajendra Akerkar, Technomathematics Research Foundation, India Rodrigo Roman, University of Malaga, Spain Said Assar,Institut TELECOM Sud Paris, France Shoba Tegginmath, Auckland University of Technology, New Zealand Shukor Abd Razak, Universiti Teknologi Malaysia, Malaysia Sokratis K. Katsikas, University of Piraeus, Greece Spiros Sirmakessis, University of Patras, Greece Spyros Kokolakis, University of the Aegean, Greece Steven Furnell, University of Plymouth, UK Susanne Robra-Bissantz, University of Braunschweig, Germany Tadashi Nakano, University of California, USA Thanassis Tiropanis, University of Southampton, UK Tihomir Vraneševic, Faculty of Economics and Business, Croatia Tokuro Matsuo, Yamagata University, Japan Yannis Charalabidis, National Technical University of Athens, Greece Yanwei Pang, Tianjin University, China Yingjie Yang, De Montfort University, United Kingdom Zeljko Panian, University of Zagreb, Croatia xiii xiv KEYNOTE LECTURE ELECTRONIC WORD-OF-MOUTH ON WIRED VERSUS WIRELESS INTERNET: HOW CAN WE UNDERSTAND SOCIAL INFLUENCE THEORY IN MOBILE COMMERCE? Dr. Shintaro Okazaki Department of Finance & Marketing Research College of Economics & Business Administration Universidad Autónoma de Madrid Spain This presentation addresses a phenomenon of electronic word-of-mouth in two different contexts: PC Internet and mobile Internet. Compared with laptop or desktop computer, mobile device offers greater flexibility in time and space, thus enabling consumers to be connected online more continually. In addition, small size, portability, and ease of use with location-based capabilities facilitate sending and receiving timely information in the right place. Drawing upon a social influence model proposed by Dholakia et al. (2004), I propose a causal model for consumer participation in electronic word-of-mouth (eWOM), and compare the effects of PC-based and mobile-based eWOM (hereafter pcWOM and mWOM, respectively). I posit social identity, motivations (purposive value, social enhancement, and intrinsic enjoyment), inherent novelty seeking, and opinion leadership as antecedents affecting desire (individual-level driver) and social intention (group-level driver) to engage in eWOM. I collected 271 survey responses from consumers in Japan. The proposed model fits the data reasonably well; all hypotheses are supported. Our results reveal that desire only partially mediates the effects on social intention of social identity. Compared with pcWOM participants, mWOM participants exhibit significantly higher perceptions on social intention, intrinsic enjoyment, and cognitive social identity. In general, the perceived levels of constructs were found to be higher in mWOM than in pcWOM, with regard to social intention, desire, purposive value, intrinsic enjoyment, cognitive social identity, and affective social identity. Statistically, the difference was most striking in social intention, intrinsic enjoyment, and cognitive social identity, all of which are likely to be perceived more favorably by participants in mWOM than by participants in pcWOM. One possible explanation for this may be that consumers who exchange information through mobile device tend to be more conscious, and more intentional, than those who exchange information through PC. After all, a mobile device is a ‘telephone’, the primary objective of which is message transmission, while a PC is a ‘processor’, with the primary objective of data transmission. Taken together, the mobile is indeed a better communication medium than the PC, and this motivates consumers to be more active in WOM. This is probably one of the most important differences between mWOM and pcWOM. xv xvi Full Papers IADIS International Conference e-Commerce 2009 BARRIERS OF ONLINE SHOPPING IN DEVELOPING COUNTRIES: CASE STUDY IRAN Omid Bigdeli; Sanam Afaghzadeh; Amir Albadvi; Neda Abdolvand IE Dept., Engineering Faculty, Tarbiat Modares University ABSTRACT While a large number of customers in the USA, Canada and European countries shop on the Internet over and over again, online shopping in developing countries is in the infancy stage. Therefore, in this research we want to explore the reasons that shoppers do not intend to purchase online in developing countries. By applying the Theory of Planned Behavior (TPB) we want to scrutinize the impediments of Internet shopping in general and Internet grocery shopping in particular. We pin point the factors that affect online grocery shopping by analyzing the data. We notice that among the 10 factors that influence Internet grocery shopping; social, technical, confidence, trust and affection are of main importance for customers in developing countries such as Iran. KEYWORDS Online shopping, Internet, Theory of Planned Behavior (TPB), grocery shopping, electronic commerce 1. INTRODUCTION By the accretion of Internet accessibility, it is estimated that in future huge amount of transactions in all over the world will be done through the Web. This probability will bolster when people deliberate about the nuts and bolts of online shopping. By considering a number of advantages like time saving, avoidance of driving, picking and packing processes willy-nilly the economy tends to explicit from these benefits. On this ground, Chan et al. (2002) rest on the belief that while electronic commerce still constitutes a diminutive amount of many countries' economies, it is seen as an opportunity to decrease cost and improve efficiency in other countries (Chanand Al-Hawamdeh 2002). Despite the fact that online shopping has lots of advantages, when companies decide to establish it, they have some problems in their business like employee’s resistance, eliminating some departments and changing some structures. Of course, these are not all difficulties. After taking away of all the company’s inhibitors, people resist to do shopping online. Two of the most widely used psychological theories which concentrate on the link between attitude and behavior are Technology Acceptance Model (TAM) proposed by Davis (1989) and Theory of Planned Behavior (TPB) offered by Icek Ajzen (1991). These two models have both been adapted from Fishbein and Ajzen's (1980) Theory of Reasoned Action (TRA). TAM explains and predicts individual's acceptance of a technology. On the other hand, TPB is one of the most predictive persuasion theories being applied to discover the relations among beliefs, attitudes, behavioral intentions and behaviors in various fields (Ramus and Nielsen 2005). This theory has been widely used in analyzing consumer behavior (East 1993, Conner 1993, Taylor and Todd 1995, Dennison and Shepherd 1995, Thompson and Thompson 1996, Bredahl and Grunert 1997, Povey and Conner 2000, Bredahl 2001, Scholderer and Grunert 2001). Since TPB is a link between attitude and behavior, it is a more promising theoretical framework for an in depth exploration of beliefs and barriers held by consumers' Internet grocery shopping in Iran. By studying the literature we ascertain the factors which affect customers’ decision to purchase online compared with conventional grocery shopping in terms of convenience, product range, price, the risk of receiving inferior quality groceries and the loss of the recreational aspect of grocery shopping. This conceptual paper tends to decipher what parameters explain consumers’ willingness to buy groceries on the Internet. It starts with an introduction, continued with a brief explanation about the theory of planned behavior and then followed by the obstructions. In the third section, the methodology of this research has 3 ISBN: 978-972-8924-89-8 © 2009 IADIS been illustrated. It continues by analysis of the data and at the end of this study, we express momentous hints which explicit from our data. A variety of studies have attempted to profile Internet shoppers, mainly with regard to demographic and, to a lesser extent, psychographic criteria which is given in the table1. Table 1. The aim of the research in literature review in brief Aim of Research E-commerce developing countries Characterize shoppers Factors customers’ making Authors in Ogawara S., Chen J.C.H. et al. 2003; Kshetri, 2005; Molla A. and Licker P.S., 2005; Lin H.F.,2007; Moon J., Chadee D., et al. , 2008 internet Bellman et al., 1999; Weber and Roehl, 1999; Choi J. and Leung K. 2003; Colley A. and Maltby J., 2008; Childers T.L. and Scarborough C.K., 2008 George, 2004; Choi J. and Geistfeld L, 2004; Garbarino E. and Strahilevitz M., 2004; Chen Y.H. and Barnes S. 2007; Connolly R. and Bannister F. 2008; Ha S. and Stoel L., 2008 Effecting decision Consumers attitude toward online shopping Monsuwe´ et al., 2004; Ramus and Nielsen, 2005; Hansen T. 2005; May So W.C., Danny Wong T.N. et al. 2005; Bridges E. and Florsheim r., 2008 Shoppers intention purchase online Donthu and Garcia, 1999; Morganosky and Cude, 2000; Raijas and Tuunainen, 2001; Jayawardhena, 2004; Huang Y. and Oppewal H., 2006 to 2. A REVIEW ON THEORY OF PLANNED BEHAVIOR (TPB) One of the most widely used social psychology theories about the way in which perceptions influence actions is Ajzen’s Theory of Planned Behavior (Ajzen 1991, East 1993, Conner 1993, Taylor and Todd 1995, Dennison and Shepherd 1995, Thompson and Thompson 1996, Bredahl and Grunert 1997, Povey and Conner 2000, Limayem and Khalifa 2000, Bredahl 2001, Scholderer and Grunert 2001). As mentioned before in psychology, this theory is concerning the link between attitudes and behavior. Figure 1 is an illustration of TPB. Figure 1. Theory of Planned Behavior The three determinants of TPB are: Outcome Beliefs (Attitude), Normative Beliefs (Subjective Norms) and Control Beliefs (Perceived Control). The valence of the action for the potential actor can be referred as attitude, which is determined by outcome beliefs (person’s beliefs about the expected outcome of a given behavior). Perceived social pressure, to either perform an action or not, is known as subjective or social norms determined by normative beliefs. One of the factors important in formation of intention is ease of use; this can be specified by perceived behavioral control. Beliefs about factors that facilitate or impede the performance of the behavior, control beliefs, are ascertaining perceived behavioral control (Kotler 2002, Ramus and Nielsen 2005, Shih 2008, Vermeir and Verbeke 2008). In this exploratory study of consumers’ perceptions about Internet grocery shopping, the TPB provides a systematic, comprehensive account of psychological factors with a potential power to explain why people use or what do they abstain from using the Internet for shopping groceries. The factors are as below: • Convenience of shopping: 4 IADIS International Conference e-Commerce 2009 Over the history of online shopping, convenience is the most critical parameter in the mind of customers (GVU 1998, Wolhandler 1999, Wolfinbarger and Gilly 2001, Berry and Seiders 2002, Grewal and Iyer 2002, Raijas 2002, Forsythe and Shi 2003, Monsuwe and Dellaert 2004, Huang and Oppewal 2006) • Enjoyment and fun of shopping: Shopping enjoyment is the pleasure one gains from the buying procedure (Beatty and Ferrell 1998). It returns to the differences between hedonic and utilitarian. As the same time as utilitarian behave shopping as a work, hedonic people strive to gain fun and entertainment in shopping process (Babin and Darden 1994). Meanwhile Childers et al. (2001) found “enjoyment” to be a consistent and strong predictor of attitude toward online shopping. If consumers get pleasure from online shopping experience, they have a more positive attitude toward online shopping (Childers and Carr 2001). • Technical system and web design: By considering the nuts and bolts of online shopping, it would not be wrong to conclude that this kind of shopping has some difficulties. As a case in point, many households can not apply the Internet for the shopping process and also do not access to qualified Internet connection (Hammond 2001, ONS 2002,Huang and Oppewal 2006). Moreover, there exists some websites requiring long time to upload (GVU 1998, Forsythe and Shi 2003). Figure 2. Obstructions in Online Shopping • Range of available products: By considering the recent surveys it can be revealed by utilizing online shopping, purchasers have access to wider range of store and products (GVU 1998, Huang and Oppewal 2006, Shih 2008). • Price, bargains, costs: Online shoppers can easily evaluate the products specifications and prices more efficiently and effectively than offline stores (Monsuwe and Dellaert 2004). Meanwhile you should keep in mind that consumers assume that delivery fee is an additional cost (Hammond 2001, Ramus and Nielsen 2005). • Social aspect of shopping: Many researchers hold on the opinion that social interaction is a missing parameter in online shopping (Ramus and Nielsen 2005). • Personal service Many people rest on the belief that lack of personal customer service in online shopping leads them to a great concern. They prefer to get some information and assistance about products at a physical location (Ramus and Nielsen 2005). • Confidence in company logistics: By considering the pros and cons of Internet shopping it can be concluded that in category of some standardized products like books, CDs, DVDs and etc. people have great positive intention to purchase them via the Internet, in contrary; they abstain from buying some products like vegetable, fruits, meats and perfumes inasmuch as they desire to see, feel, touch and smell them (Watson and Berthon 2000, Hammond 2001, Raijas 2002, Forsythe and Shi 2003, Monsuwe and Dellaert 2004, Ramus and Nielsen 2005). Although because of the separation of purchase and delivery process there is less assurance in receiving the products on time (Hammond 2001, Ramus and Nielsen 2005). 5 ISBN: 978-972-8924-89-8 © 2009 IADIS • General trust in company: One of the most significant reasons not to purchase via Internet is trust. The level of security and privacy has influenced on the online shoppers, it means that high level of security has a great positive effort on consumers' attitude toward Internet shopping (Rotter 1971, Jacobs 1997, Lee and Turban 2001, McKnight and Chervany 2001-2002, Grewal and Iyer 2002, Forsythe and Shi 2003, Choi and Leung 2003, Monsuwe and Dellaert 2004, Hammond 2001). • Influence of colleagues, family and friends: According to the research conducted by Ramus and Nielsen (2005), a large number of people recommend to actuate family, friends or colleagues to take up Internet shopping. • Exposure to advertisements and accidental browsing across interesting commercial web sites: According to the research conducted by Ramus and Nielsen (2005), they mentioned that these two recent factors affect on consumers' intention toward online shopping. 3. METHODOLOGY In order to figuring out the impediments and their importance, in accord with our model, we designed a 54 item questionnaire. Then, we test the questionnaire among 12 people and since the Cronbach’s alpha was greater than 0.7, it is valid for distribution. The questionnaire contains some personal questions about gender, age; marital status and the rest of the questions were categorized in 10 groups each representing a factor of the model. We have distributed the questionnaire among 307 persons through the e-mail and face to face. We received 275 completed questionnaires; it was about 89 percent return back rate. 4. ANALYSIS We analyze the questionnaires and run a descriptive analysis on the answers. Results are depicted in table 2. Table 2. Discriptive Analysis Disagreed Neutral Agreed Social 12.36 11.73 75.91 Technical 5.42 15.49 79.09 Confidence 4.36 27.09 68.55 Trust 15.45 3.82 80.73 8.73 18 73.27 Affection • Social We found that the social factors are important for the participants. In retrospect and over the previous literature we can found that lack of interaction is an important hurdle for online shopping (Steenkamp et al.1999, Rosen and Howard 2000, Grewal and Iyer 2002, Choi and Leung 2003, Arnold and Reynolds 2003, Ramus and Nielsen 2005, Huang and Oppewal 2006). As you can see in table 2, 75.91 percent of people rest on the opinion that Internet grocery shopping lacks social aspects in comparison with conventional purchasing. In eastern countries like Iran, many people go shopping since they can interact with other people. Figure 3. Social 6 IADIS International Conference e-Commerce 2009 • Technical The speed of Internet connection is vital for the users (http:// www.pressroom.com/~screenager/broadband/Intro.html, Kshetri 2007). Since most of people use dial up connection in Iran, so page loading is time consuming; on the other hand many participants mentioned that WebPages are befuddling and confusing. Therefore, Iranian people do not sense the usefulness of Internet shopping. Figure 4. Technical • Confidence Confidence of shopping returns to some issues like: logistics, delivering on time and intact products. It was mentioned through the history of online shopping researches and there exists both in developing countries and developed countries (Ramus and Nielsen 2005, Huang and Oppewal 2006, Vermeir and Verbeke 2008). As a case in point, Tesco.com in England removes this matter by applying IT systems and trained staffs as pickers; every picker cover a distinct area in the store, they have a PDA and the items have sent to their PDA electronically and the pickers collect the items through the shelves attentively and they check the quality and expire dates as if they are shopping for themselves (Jelassi and Enders 2005). Figure 5. Confidence • Trust Trust is one of the most significant problems in e-commerce (Lee and Turban 2001, Grewal and Iyer 2002, O’Cass and Fenech 2003, Monsuwe and Dellaert 2004, Ramus and Nielsen 2005, Kshetri 2007). Over the history, eastern people used to transact face to face (Kshetri 2007), but recently by the growth of electronic banking in Iran and benefiting from its advantages, people tend to utilize it. • Figure 6. Trust Affection Based on individualism-collectivism axiom (Choi and Leung 2003) it was unsurprising that 73.28 percent of people said that they are affected by their friends, colleagues and family. In fact, Iranian people used to counsel about their activities by their friends and since the general overview is negative to online shopping, people are not eager to use it. Figure 7. Affection 5. DISCUSSION AND CONCLUSION As we cited in the introduction, since the online shopping in developing countries like Iran is in formative years and the customers are not professional yet; it can be said that managers in these countries have easier mission to absorb the potential customers to shop online in comparison with the customers in developed countries. By perusing the previous literature in the field of electronic commerce, online shopping, Internet grocery shopping and the results of this research; we could develop the aforementioned model about the inhibitor factors of online grocery shopping. According to our results the entire parameters break down to three clusters and each group has a number of sub-factors. Some of these barriers such as social interaction, technical problems, security and trust have been mentioned over and over again through the literature. Social, technical and affection problems are environmental factors which companies cannot affect robustly on them, but companies can improve logistics and trust issues by innovative use of IT and by cooperating with other eminent organizations and banks. Seeing that we have some failed cases in online shopping in Iran, it seems that new companies should consider the previous cases to remove the former 7 ISBN: 978-972-8924-89-8 © 2009 IADIS errors. As an illustration, the companies' websites were not updated and also most of them have some problems about the delivering process. Also, customers suffer from incomplete menu of the products. For instance, in most cases the websites merely offer a special brand and buyers cannot access to their favorite brands; also the products of the offered brand are not complete and the customers only access to inadequate number of goods. Generally, it can be said that by considering some environmental issues such as terrible traffic jam in Tehran and restricting use of petrol and on the other hand by the accretion of using Internet among Iranian people; Internet grocery shopping willy-nilly will be utilized and pioneers can benefit from investing in this market. REFERENCES Ajzen I. (1991). "The theory of planned behavior: Some unresolved issues." Organizational Behavior and Human Decision Processes, 50: 179–211. Arnold M.J., Reynolds k.E., (2003), “Hedonic shopping motivations”, Journal of retailing, 79, 77-95 Babin B. J., Darden W.R. (1994). "Work and/or fun: measuring hedonic and utilitarian shopping value." Journal of Consumer Research 20(4): 644–656. Beatty S.E. and Ferrell M.E. (1998). "Impulse buying: modeling its precursors." Journal of Retailing 74(2): 169-91. Bellman S., Lohse G.L. (1999). "Predictors of online buying behavior." Communications of the ACM 42(12): 32-8. Berry L. and Seiders K. (2002). "Understanding Service Convenience." The Journal of Marketing 66(3): 1-17. Bredahl L. (2001). "Determinants of consumer attitudes and purchase intentions with regard to genetically modified foods: results of a cross-national survey." Journal of Consumer Policy 24: 23-61. Bredahl L. and Grunert K.G. (1997). "Determinants of the consumption of fish and shellfish in Denmark: an application of the theory of planned behavior." Seafood from Producer to Consumer: Integrated Approach to Quality, Elsevier, Amsterdam: 21-30. Bridges E. and Florsheim r. (2008). "Hedonic and utilitarian shopping goals: The online experience." Journal of Business Research 61: 309–314. Chan B. and Al-Hawamdeh S. (2002). "The development of e-commerce in Singapore, The impact of government initiaives." Business Process Management Journal, 8(3): 278-288 Chen Y.H. and Barnes S. (2007). "Initial trust and online buyer behavior." Industrial Management & Data Systems 107(1): 21-36. Childers T.L. and Scarborough C.K. (2008). "Expanding opportunities for online shoppers with disabilities." Journal of Business Research(doi:10.1016/j.jbusres.2008.06.017). Childers T.L., Carr C.L. (2001). "Hedonic and utilitarian motivations for online retail shopping behavior." Journal of Retailing 77(4): 511-35. Choi J. and Geistfeld L. (2004). "A cross-cultural investigation of consumer e-shopping adoption." Journal of Economic Psychology 25: 821–838. Choi J. and Leung K. (2003). "Risk perceptin and e-shopping: a cross-cultural study." Journal of Fashion Marketing and Management 7(1): 49-64. Colley A. and Maltby J. (2008). "Impact of the Internet on our lives: Male and female personal perspectives." Computers in Human Behavior 24: 2005-2013. Conner M.T. (1993). "Understanding determinants of food choice." British Food Journal 95(9): 27-31. Connolly R. and Bannister F. (2008). "Factors influencing Irish consumers’ trust in internet shopping." Management Research News 31(5): 339-358. Dennison C.M. and Shepherd R. (1995). "Adolescent food choice: an application of the theory of planned behavior." Journal of Human Nutrition and Dietetics 8(9-23). Donthu N. and Garcia A. (1999). "The internet shopper." Journal of Advertising Research 39(3): 52–8. Forsythe S.M. and Shi B. (2003). "Consumer patronage and risk perceptions in Internet shopping." Journal of Business Research 56: 867– 875. Frontline.net, Broadband in the Developing World, 2001. http:// www.pressroom.com/~screenager/broadband/Intro.html (retrieved 21.04.01 Garbarinoa E. and Strahilevitz M. (2004). "Gender differences in the perceived risk of buying online and the effects of receiving a site recommendation." Journal of Business Research 57: 768– 775. 8 IADIS International Conference e-Commerce 2009 George J.F. (2004). "The theory of planned behavior and internet purchasing." Journal of Internet Research 14(3): 198212. Grewal D., Iyer G.R. (2002). "Internet retailing: enablers, limiters and market consequences." Journal of Business Research. GVU’s 9th. (April–May 1998.). WWW User Surveys. Ha S. and Stoel L. (2008). "Consumer e-shopping acceptance: Antecedents in a technology acceptance model." Journal of Business Research (In Press). Hammond A. (2001). "Digitally Empowered Development." Foreign Affairs 80(2): 96-106. Hammond K. (2001). "B2C e-commerce 2000-2010: what expert predict." Business Strategy Review 12(43-50). Hansen T. (2005). "Consumer adoption of online grocery buying: a discriminate analysis." International Journal of Retail & Distribution Management 33(2): 101-121. Huang Y. and Oppewal H. (2006). "Why consumers hesitate to shop online An experimental choice analysis of grocery shopping and the role of delivery fees." International Journal of Retail & Distribution Management 34(4/5): 334-353. J. R. East, (1993) “Investment decisions and the theory of planning behavior,” J. Econ. Psych., vol. 14, pp. 337–375. Jacobs P. (1997) Privacy: what you need to know. InfoWorld; 19(44):111 – 2. Jayawardhena C. (2004). "Personal values’ influence on e-shopping attitude and behavior." Internet Research 14(2): 12738 Jelassi T. and Enders A. (2005). “Strategies for e-business”, Prentice Hall, ISBN: 0 273 688405 Kotler. P (2002). Marketing Management. Boston, Prentice-Hall, Inc. Kshetri, N. (2007). "Barriers to e-commerce and competitive business models in developing countries: A case study." Electronic Commerce Research and Applications. Lee M.K.O. and Turban E. (2001). "A trust model for consumer internet shopping." International Journal of Electronic Commerce 6(1): 75-91. Limayem, M., M. Khalifa (2000). "What makes consumers buy from the internet? A longitudinal study of online shopping." Systems and Humans 30(4): 421-32. Lin H.F. (2007). "Predicting consumer intentions to shop online: An empirical test of competing theories." Electronic Commerce Research and Applications 6: 433-442. May So W.C., Danny Wong T.N. (2005). "Factors affecting intentions to purchase via the internet." Industrial Management & Data Systems 105(9): 1225-1244. McKnight D.H., and Chervany N.L., (2001-2002)., “What trust means in e-commerce customer relationships: an interdisciplinary conceptual typology”, International Journal of Electronic Commerce, Vol. 6 No. 2, pp. 35-59. Molla A., Licker P.S., (2005) Perceived e-readiness factors in ecommerce adoption: an empirical investigation in a developing country, International Journal of Electronic Commerce 10 (1): 83–110. Monsuwe T. and Dellaert B. (2004). "What drives consumers to shop online? A literature review." International Journal of Service Industry Management 15(1): 102-121. Moon J., Chadee D. (2008). "Culture, product type, and price influences on consumer purchase intention to buy personalized products online." Journal of Business Research 61: 31-39. Morganosky M.A. and Cude B.J. (2000). "Consumer response to online grocery shopping." International Journal of Retail Distribution Management 28(1): 17-26. O’Cass A., Fenech T., (2003),”Webretailing adoption: exploring the nature of internet users Webretailing behavior “, Journal of retailing and consumer services, 10, 81-94 Ogawara S., Chen J.C.H. (2003). "Internet grocery business in Japan: Current business models and future trends." Industrial Management & Data Systems 103(9): 727-735. ONS (2002). "Internet access: household and individuals." Office for National Statistics, London. Povey R.and Conner M. (2000). "Application of the theory of planned behaviour to two dietary behaviors: roles of perceived control and self-efficacy." Journal of Health Psychology 5: 121-39. Raijas A. (2002). "The consumer benefits and problems in the electronic grocery store." Journal of Retailing and Consumer Services 9: 107–113. Ramus K. and Nielsen N. A. (2005). "Online grocery retailing: what do consumers think?" Internet Research 15(3): 335352. Rosen, K. T., & Howard, A. L. (2000). “E-retail: gold rush or fool’s gold?”, California Management Review, 42 (Spring), 72–100. Rotter, J.B. (1971), “Generalized expectancies for interpersonal trust”, American Psychologist, Vol. 26 No. 5, pp. 443-52 Scholderer J. and Grunert K.G. (2001). "Does generic advertising work? A systematic evaluation of the Danish campaign for fresh fish Aquaculture." Economics and Management 5(5/6): 253-72. 9 ISBN: 978-972-8924-89-8 © 2009 IADIS Shih H.P. (2008). "Using a cognition-motivation-control view to assess the adoption intention for Web-based learning." Computers & Education 50: 327–337. Steenkamp, J. E. M., Hofstede, F., & Wedel, M. (1999). A cross-national investigation into the individual and national cultural antecedents of consumer innovativeness. Journal of Marketing, 63, 55–69 Taylor S. and Todd P.A. (1995). "Understanding information technology usage: a test of competing models." Information Systems Research 6(2): 144-176. Thompson N.J. and Thompson K.E. (1996). "Reasoned action theory: an application to alcohol-free beer." Journal of Marketing Practice 2(2): 35-48. Vermeir I. and Verbeke W. (2008). "Sustainable food consumption among young adults in Belgium: Theory of planned behaviour and the role of confidence and values." Ecological Economics 64: 542-553. Watson R.T., Berthon P. (2000). Electronic Commerce, Mike Roche. Weber K. and Roehl W.S. (1999). "Profiling people searching for and purchasing travel products on the world wide web." Journal of Travel Research 37: 291-8. Wolfinbarger M. and Gilly M.C. (2001). "Shopping online for freedom, control, and fun." California Management Review 43(2): 34-55. Wolhandler HC. (May 1999.). Real numbers behind ’Net Profits 1999 (6th annual survey of online commerce). 10 IADIS International Conference e-Commerce 2009 FACTORS AFFECTING CONSUMERS ADOPTION OF ECOMMERCE IN SAUDI ARABIA FROM A CONSUMERS’ PERSPECTIVE Fahad Aleid, Simon Rogerson and Ben Fairweather Faculty of Technology De Montfort University Leicester, UK ABSTRACT This paper presents a literature review on the global barriers to consumers’ adoption of ecommerce. It also shows the state of ecommerce in Saudi Arabia and goes on to identify the factors that affect consumers’ adoption of ecommerce in Saudi Arabia from their perspective. KEYWORDS Ecommerce, adoption, qualitative research, grounded theory. 1. INTRODUCTION Ecommerce related to consumers has grown enormously (Ranganathan and Ganapathy, 2002). However, while ecommerce has grown significantly in the first world, it lags in developing countries. Research using surveys and the Theory of Planned Behaviour (TPB) mention that consumers' attitudes are one of the barriers to ecommerce growth (Molla and Licker, 2005 and Sait et al. 2004). Moreover, research indicates that businesses in developing countries will face more risks than those in developed ones (Molla and Licker, 2005 and Molla and Licker, 2005a). The study of ecommerce systems differ from country to country according to culture, human behaviour and national infra-structure (Barbonis and Laspita, 2005) and other issues that will become evident from this study. Saudi Arabia is one of these developing countries. It has shown an interest in adopting all aspects of technology for its economic activities by building appropriate infra-structures for egovernment and ecommerce (Sait et al., 2004 and Al-Shehry et al., 2006). However, are consumers satisfied with ecommerce in Saudi Arabia and will they adopt this new technology? Few studies relate to ecommerce in developing countries and Saudi Arabia in particular. Those studies that exist indicate that consumer attitude is likely to be one of the main barriers that restrict ecommerce growth. However, these studies did not investigate the factors that affect its adoption or the barriers that prevent consumers from adopting it (Molla and Licker, 2005, Sait et al., 2004, Molla and Licker, 2005a and Al-Shehry et al., 2006). Therefore, this study addresses this gap regarding ecommerce in Saudi Arabia in general and the consumer’s acceptance of ecommerce in particular. This study will give insights into consumer acceptance of ecommerce in Saudi Arabia. 2. ECOMMERCE STAGE IN SAUDI ARABIA Saudi Arabia’s population is 24 million spread over an area of 2.25 million square kilometres (World Bank, 2006). According to Internetworldstats.com, Saudi Arabia have 6.2 million Internet users as of March 2008. Still Saudi Arabia has no national plan for ecommerce. Companies in Saudi Arabia can be divided into four different categories according to their stage in ecommerce. The first category is the large companies such as airlines, the banking sector and telecommunication companies. These companies have their own websites and they sell, serve, advertise and deliver online. For example, airlines in Saudi Arabia allow the consumers to 11 ISBN: 978-972-8924-89-8 © 2009 IADIS buy e-tickets, pay online by credit cards and then send the e-tickets to the consumers’ email. Telecommunication companies’ websites allow their customers to check their statements and to pay online. However, consumers collect their purchased items at a shop. The bank sector also gives their consumers full online account services, such as, online statements, transferring money, selling and buying stocks etc. and then sends the monthly statements through the Saudi Post. The majority of Saudi banks request from their customers to collect their debit or credit cards at branches or at a private shipping company branches. This is because of the lack of clarity about home addresses in Saudi Arabia. The second category is shops that have an online website offering information, addresses and catalogues with prices. This group offers no online transactions. A good example is Jarir Bookstore (jarirbookstore.com) which has approximately 20 bookstores over Saudi Arabia. The Third category is shops that just have a basic website that offers information such as addresses, locations and contacts such as Extra shops (extra1.com), which has branches across the Kingdom. The fourth category, the most common do not have websites. Therefore, in Saudi Arabia, there is a variety of ecommerce stages. Some companies are at an advanced stage while others are not involved in ecommerce with no plans for future involvement. 3. LITERATURE REVIEW In the literature, there are reports of barriers to consumer adoption of ecommerce that are globally recognised. The reader can then compare these global barriers with barriers in Saudi Arabia that relate to the consumers’ perspectives. Barbonis and Laspita (2005, p31) stated that lack of trust is one of the barriers to ecommerce adoption. Karakaya and Charlton (2001, p49) declared that bandwidth (high-speed Internet), privacy and security of transactions are key barriers for consumers. Moreover, they mentioned that product delivery and return are also barriers. Farhoomand et al (2000, p27) stated in their research findings that there are six barriers to global ecommerce. These are the legal barrier (which includes user privacy, copyright laws, legitimacy of electronic signatures, ISPs’ responsibilities, cryptography and public key encryption). Second is the technical barrier (which includes security, infrastructure, and the integration of old systems and the availability of applications that can support local languages). The third is the economic barrier (which includes income, literacy level, social infrastructure, currency and the cost of buying technologies). The fourth is the cultural barrier (which includes ethnicity, religion and language). The fifth is the organizational barrier (which includes negative attitudes, lack of knowledge, resistance to change and lack of top management commitment). The sixth is the barrier of political issues, such as, government attitudes and inter-agency coordination. Vatanasakdakul et al. (2004, p10) found that language is a very serious problem. Moreover, Vatanasakdakul et al. (2004, p7) found that many managers prefer to use a “wait and see” strategy and they are forced into ecommerce by foreign partners or by company image. Finally, they find that most of the managers are old and there is a gap between the superiors and their subordinates. Del Aguila-Obra and Padilla-Melendez (2006, p108) stated that organization is an ecommerce barrier that affects Internet technology adoption. Chirch and Kauffman (2000, p66) defined conversion barriers that are found in converting from the old system to the new one and include resource, knowledge and usage barriers. Barboonis and Laspita (2005, p31) mentioned security, privacy, the state of anonymity of the Internet, restricted ways of payment and lack of personal communication as barriers to the adoption of ecommerce. Siala et al (2004, p7) declared that, “religion influences consumer purchase decisions”. Chau et al (2002, p139) found that culture influences consumers’ attitudes towards ecommerce. Similarly, Barbonis and Laspita (2005, p33) stated that cultural factors, personal characteristics (attitudes to trust and risk) and demographic factors are some of the factors that influence adoption in Greece. 4. RESEARCH METHODOLOGY The research uses grounded theory (GT) to carry out qualitative research following Strauss and Corbin’s (1990) approach. Strauss and Corbin (1990, p24) explained the use of literature in grounded theory and pointed out that a literature review is a good background and can be used in the research. However, they pointed out that the researcher does not need to review all the literature about his study. In the field, semistructured interviews were conducted. Subsequently questions were developed after each interview based on 12 IADIS International Conference e-Commerce 2009 the interviewees’ answers. The researcher continued to conduct interviews until the stage when interviewees were unable to elicit any further useful information. Twenty-two interviews were conducted with interviewees selected according to gender, age, income, education and geographical area. Table 1 shows the distribution in detail. Table 1. Demographic Spread of the Interviewees Conducted Number of interviewees North 4 Saudi Arabian regions South Middle East 4 6 4 West 4 Gender Male Female 2 2 2 2 4 2 2 2 2 2 Age 15-24 25-34 35-54 >=55 1 2 1 1 2 1 3 2 3 1 1 2 1 1 1 2 2 4 1 3 1 1 2 1 1 2 1 1 1 1 1 Education Low Medium High 1 1 3 Income <4000 4000-7999 8000-12000 >12000 2 2 1 1 1 1 1 1 3 1 Living in Village Town City 1 1 2 1 1 2 1 5 3 1 1 2 1 1 2 1 1 3 2 1 1 1 1 1 1 1 1 1 Employment status Unemployed Government employee Private sector employee Retired Self employed Student 1 1 5. RESEARCH FINDINGS After transcribing and analyzing the interviews, the following factors were found to affect consumer adoption of ecommerce in Saudi Arabia from consumers’ perspectives: 5.1 The Internet Infrastructure 5.1.1 The Availability of Broadband Internet infrastructure has improved in most Saudi cities and DSL has come to be a good alternative for most citizens. However, villages and small towns in Saudi Arabia are still at a disadvantage with only dial up Internet connections. One of the interviewees from the northern region of Saudi Arabia said, “Most villages in the northern region don’t have the internet infrastructure. They do not have broadband. They are still using dial up”. In addition, an interviewee from a small town in the middle of the country said, “There are no DSL services in our town. We are still surfing the Internet using dial up”. However, people in the main cities are still suffering from the Internet disconnecting from time to time. According to an interviewee who has broadband and is living in a main city “Sometimes, the Internet is disconnected for hours or even for a whole 13 ISBN: 978-972-8924-89-8 © 2009 IADIS day”. In addition, there is a waiting list for broadband in some of the main cities: “You have to wait for a long time to get a broadband connection”. 5.1.2 The Internet Cost People who live in the main cities on high incomes said that the monthly charge for the Internet is fair. In addition, they stated that Internet prices have reduced. However, people who are living in cities on low incomes, small towns and villages stated that Internet prices were still expensive. For example, an interviewee from the northern region said, “The Internet price is still expensive. I do not think that my parents are able to pay the broadband’s monthly cost”. However, two citizens from one of the main cities in Saudi Arabia said, “The Internet cost is cheaper than before” and “However, comparing with other countries, it is really expensive”. Table 2 shows Saudi consumers and their point view. It also indicates that Internet prices in Saudi Arabia and some developing countries are still expensive in comparison to the UK. Table 2. Internet Service Prices in January 2009 Sources (Amf.org.ae, dol.gov, du.ae, etisalat.ae, orange.jo) Country KSA UAE Jordan Egypt UK Internet Speed 128KB 256KB 512KB 1 MB 2 MB 8 MB 16 MB £19 N/A £16 N/A N/A £27 £27 N/A £11 N/A £36 £34 £20 £18 N/A £54 £45 £28 £31 N/A £83 £63 £37 £50 £4.5 or free £152 £109 £56 £105 £4.5 £174 N/A N/A £186 £10 Gross Domestic Product Per Capita £=5.3SR £9,900 £26,000 £1,700 £1,000 £21,000 5.1.3 Internet Speeds The Internet speed in Saudi Arabia reaches 20MB but unfortunately, not everyone can get the speed they require because of the high price of Internet subscriptions. Consumers may not get the speed for which they subscribed due to the proxy limitations in Saudi Arabia. According to an interviewee who lives in a city in the eastern region, “The Internet speed is slow even if you order a high speed connection. We feel that there is a high load on the Internet”. The slow speed of the Internet makes surfing boring and this will make consumers reluctant to purchase through the Internet. An interviewee from the eastern region stated that, “Sometimes, I cannot work on the Internet and I turn off my computer because the Internet is slow and I feel bored to death”. 5.2 The Official Postal Carrier, Delivery and Return Process The official carrier in Saudi Arabia is the Saudi Post. However, most of the interviewees stated that there is a lack of trust in the services. An interviewee from the eastern region stated that, “Saudi Post is not doing its job that well; however, other private post companies are covering the gap for much more money”. Another one from the northern region stated, “My family and other families are sharing the same post box. One member of one of the other family brings our post every two or three weeks. So I cannot order private items through this post box”. Another interviewee from the southern region stated, “The material might be late, lost or damaged” and “If I am not satisfied with the purchased item, how to return it back, especially if it’s from outside my city. I cannot trust the main post carrier and I am not sure that the shop will accept the returned item”. 5.3 The Lack of Co-operation between Online Shops and the Postal Carriers and the Lack of Knowledge about the Terms and Conditions According to some of the interviewees, the Saudi Post does not check the item’s status before delivery. They deliver the item even if it is broken, without mentioning this to the consumer. Furthermore, there is a lack of knowledge of the terms and conditions of the Saudi Post. There is no link to the terms and conditions in their website. Consumers and shops in Saudi Arabia have no idea about the main carrier’s responsibilities. Two 14 IADIS International Conference e-Commerce 2009 citizens from the eastern region stated these issues by saying, “So what will happen if you order online and receive a broken item. The Saudi Post and the online shop will not return it back and no one will be in charge. I prefer to see the item myself and check its status”. 5.4 The Absence of the Postal Addresses Systems for Homes and Shops in Saudi Arabia Saudi Arabia does not have clear addresses for homes and shops. This makes the delivery process more complicated. Designing a postal addresses system in Saudi Arabia would help to expand ecommerce and business since business organizations face difficulties in locating homes and consumers face difficulties in mapping their addresses. Figure 1 shows the missing links in the ecommerce process in Saudi Arabia. Get the costumer’s order Fail Authorize the card payment Order refused Accept Create the consumer’s order Send items to customer’s address Send a confirmation email Figure 1. Simple Workflow for the Missing Link (Postal Address) in the Ecommerce Process in Saudi Arabia 5.5 Security, Risks, Privacy and Fraud Issues Some of the interviewees raised another barrier to ecommerce adoption. This barrier is related to the companies’ websites, which is the inability to distinguish between real and fake companies. An interviewee from the middle region of the country stated, “Customers in Saudi Arabia do not know about most of the Saudi websites if it’s real or fake”. In addition, an interviewee from the eastern region raised this issue, “There are many fake companies even in Saudi Arabia. These fake companies advertise and get the consumer’s money then disappear”. She continued, “Most people in Saudi Arabia had a bad experience with fake companies. So they will not trust any ecommerce website”. One of the interviewees raised a privacy issue, stating, “I think there is no law in Saudi Arabia to force firms to maintain the confidentiality of information. I cannot pass my information with no protection.” Moreover, some of the interviewees stated that there is no online protection for credit cards from having their details stolen. For example, “If my credit card is stolen, no one will protect me.” and “From my experience, Saudi banks do not protect their consumers in the online transactions”. 5.6 Ecommerce Laws, Consumer Protection Enforcement and Responsibility In 2007, the Communications and Information Technology Commission in Saudi Arabia published the Ecrimes and E-transaction Acts on the website internet.gov.sa. However, these acts seem too general and do not cover all possible e-crimes, e-transaction cases and their associated penalties. According to the interviews, consumers stated that ecommerce law is one of the main barriers to its progress. Citizens in Saudi Arabia lack knowledge of legal issues. Saudi consumers do not know whether there is an ecommerce law to protect them. According to one of the consumers who live in the northern region, “I have no idea. I have not heard about an electronic trading law in Saudi Arabia”. According to an interviewee from the middle region of the country, “There is no ecommerce law in Saudi Arabia”. Moreover, most of the interviewees do not know who should be responsible for issuing this kind of law and where to go if they experience a problem related to their online purchases. According to one of the interviewees in the middle region of the country, “If I face a problem with my online purchase, and the online shop did not solve the problem, I may not go to complain because I have no idea where to go”. 15 ISBN: 978-972-8924-89-8 © 2009 IADIS 5.7 Returns and Exchange Policies Shops in Saudi Arabia can be divided into three groups according to their returns and exchange policies. The first group and the most common are the shops that have no returns policy. The second group is the shops that represent the global brands that came from outside the country and have universal returns and exchange policies. The third group is the shops with exchange policy only. According to an interviewee from the middle region of the country, “I tried to return the purchased item back but they refused to return it back”. Another one said, “It is very difficult to return purchased items because; we do not have a returns policy”. From the interviews, it appears that there is no standard policy for the returns and exchange in Saudi Arabia. In addition, to the previous responses the reader can understand that most interviewees were afraid that they cannot return or exchange the online purchased items and this is a barrier to ecommerce consumers adopting in Saudi Arabia. 5.8 Private Business Responsibilities Saudi Arabia is suffering from a small number of online shops. According to an interviewee from the southern region, “There are some online shops. However, they are few. This is because traders were still dealing with trade the old way and do not want to change” and “This is because most of the companies are still family businesses”. The private sector has not yet taken the initiative. This is because of the old business mentality, the lack of IT professionals, a ‘wait and see’ strategy and the lack of competition. Another interviewee stated that, “This is because there is no real competition between them yet”. Therefore, customers face lack of diversity and online shops competition. 5.9 Lack of Specialized IT Companies and Professionals The Information technology field in Saudi Arabia suffers from a scarcity of IT companies and professionals. For this reason, firms are unable to meet their technical obligations in the field of electronic commerce. According to an interviewee from the south region, “Companies have not developed their systems since ages. This is because of the lack of IT professionals in Saudi Arabia”. The availability of IT companies and professionals help businesses to expand in the area of electronic commerce. It also helps to reduce the cost of designing and managing online shops. 5.10 Lack of Interest in Educating People and Society on Electronic Commerce People should be made aware of the new technology, its benefits and the best way to negotiate it. Interviewees stated that citizens should be educated about ecommerce. This can help to overcome barriers between citizens and the new technology. In addition, it gives customers greater confidence in electronic shopping and helps them to adapt to it. An interviewee from the eastern region stated that, “There is a need to educate citizens about ecommerce”. Another interviewee from the middle region of the country said, “People suffer from the lack of ecommerce knowledge. They should read and learn more about ecommerce to feel confidence when purchasing items online”. According to interviewees that Saudi media, schools and other related ministries has not taken the initiative to educate people about the ecommerce advantages and disadvantages. Therefore, there will be a lack of knowledge about the ecommerce field in Saudi Arabia. 5.11 Government Responsibility The Saudi government has taken the initiative over its government programme. There is a five year plan for its development with a budget of three billion Saudi Riyals. In 2010, there should be a complete electronic government service. However, the Saudi government has not taken the initiative for electronic commerce. According to an interviewee from the middle region of the country the “Saudi government is concerned right now about egovernment, but they did not take the initiative in the ecommerce field”. 16 IADIS International Conference e-Commerce 2009 5.12 Credit and Debit Cards Most Saudi Banks promote Islamic credit cards and facilitate the process of ordering credit cards. According to our interviewees, obtaining and ordering a credit card is easy “You can order a credit card easily from the Saudi banks with a limited amount of money for example SR 2000”. However, most Saudi people prefer not to pay by credit card. An interviewee stated that this is because they are aware of the methods that credit card companies use. The majority of our interviewees prefer to pay by debit cards. They do not like credit cards, as credit card does not withdraw the payment from their bank account at the time of purchase; it combines the payments together and then withdraws the total amount from their bank account after a period of 45 days. An interviewee from the western region stated that, “I do not like to pay by credit cards because it does not withdraw the payment directly from my bank account; it withdraws the total amount in the next month. This makes me confused about my monthly expenses”. On the other hand, Saudi debit cards do not work on the Internet. According to an interviewee from the western region, “I cannot purchase products online by my debit card. It does not work online”. Figure 2 shows the missing link in the ecommerce process in Saudi Arabia that relates to debit cards. Get the consumer’s order Fail Authorize the card payment Order refused Accept Create the costumer’s order Send a confirmation email Figure 2. Simple Workflow for the Missing Link (Debit Card) in the Ecommerce Process in Saudi Arabia 5.13 Culture 5.13.1 Language Some online stores in Saudi Arabia have both Arabic and English versions, such as, Saudi Airlines. Others have the Arabic version only. In addition, none of the interviewees stated that language is a barrier. However, one of the interviewees stated, “All the online shops that I know in Saudi Arabia have the Arabic language as the default language. However, the consumer can change to the English version if he wants to”. The reader can understand that the language issue is not a barrier in the ecommerce consumers’ adoption in Saudi Arabia. 5.13.2 Religion Previous literature stated that religion might be a barrier to ecommerce. However, this research found that religion is no longer a barrier. None of the interviewees stated that religion is an ecommerce barrier. They stated that most of the Saudi banks are capable of issuing Islamic credit cards. According to an interviewee from the middle region of the country, “Alrajhi, Albilad and now Alinma Banks have setup Islamic credit cards”. This means that there is no interest on their payments. Interviewees stated that Islam religion is not a barrier to the adoption of ecommerce by consumers. They pointed out that paying interest is the only barrier. Many of the Saudis banks have solved this problem by introducing Islamic credit cards with no interest. According to Alrajhi Bank website “We have introduced a new Islamic credit card called ‘Qassit’” and “all of Alrajhi Bank’s products are fully Islamic...there are no interest charges, no late fees for delayed payments, and no hidden charges. The only charges that are applicable are the annual fees and the cash withdrawal fees”. At the present time, many Islamic scholars authorized these Islamic credit cards. They are working with the Saudi Banks to make sure they follow the Islamic process. 17 ISBN: 978-972-8924-89-8 © 2009 IADIS 6. CONCLUSION This paper has pointed out the many barriers that deter consumers from using ecommerce. These barriers have been stated from the consumers’ perspective. Some of these barriers have previously appeared in the litrature and some of them have not. There are more barriers in villages and towns than in cities. People over 55 years are less adapted to ecommerce and less interested in dealing with new technology, such as, computers. There are no differences between women and men in terms of technological trends and in terms of barriers. Religon is no more a barrier to ecommecrce adoption. The internet providers and the Saudi post should work hard to satisfy Saudi consumers. There should be more work on educating consumers about advantages, disavantages and the best way in dealing with ecommerce. However, more work should be done on how to overcome barriers to facilatate consumers in Saudi Arabia to adopt ecommerce. REFERENCES Alrajhi Bank.2008. Alrajhi Bank Credit Cards. Retrieved 21 Dec 2008. from <alrajhibank.com.sa/ Individual/Solutions /Credit Cards/Pages/cards.aspx> Al-Shehry, A. et al., 2006. The Motivations for Change towards e-government adoption: Saudi Arabia as a case study. eGovernment Workshop 06 (egov06). Brunel University, UK. AMF.2008.Broadband prices. Retrieved 20 Dec 2008. from <amf.org.ae/pages/XlsToHtmlViewer.aspx?filename=uploads /Docs/ECONOMICDEPT/Eco_Ind/INDFRM03.xls&xlsType=3> Barbonis, P.A. and Laspita, S, 2005. Some factors influencing adoption of e-commerce in Greece. Engineering Management Conference. Proceedings. 2005 IEEE International. Chau, P. et al., 2002. Cultural differences in the online behaviour of consumers. Communications of the ACM. 45(10), pp 138-143 Chirch, A. and Kauffman, R, 2000. Limits to value in electronic commerce-related IT investments. Journal of Management Information Systems. 17(2), pp 59-8 Del Aguila-Obra, A. R. and Padilla-Melendez, A., 2006. Organizational factors affecting Internet technology adoption. Internet research. 16(1), pp 94-110 Du .2008. Broadband prices. [online] Available at: www.du.ae [Accessed 28 December 2008]. Etisalat.2008. Broadband prices. [online] Available at: www.etisalat.ae [Accessed 28 December 2008]. Farhoomand, A. et al., 2000. Barriers to global electronic commerce: a cross-country study of Hong Kong and Finland. Journal of organizational computing and electronic commerce. 10(1), pp 23-48 Internetworldstats. 2009. Middle East Internet usage. [Online] Available at: http://www.internetworldstats.com/middle.htm [Accessed 28 December 2008]. Karakaya, F. and Charlton, E. T., 2001. Electronic commerce current and future practices. Managerial Finance. 27(7), pp 42-53 Molla, A. and Licker, P. S., 2005a. Perceived e-readiness factors in e-commerce adoption: an empirical investigation in a developing country." International Journal of Electronic Commerce. 10(1): 83-110. Molla, A., and Licker, P. S., 2005. E-commerce adoption in developing countries: A model and instrument." Information & Management. 42(6): 877-899. Orange.2008. Broadband prices. [online] Available at: orange.jo/adsl.php [Accessed 28 December 2008]. Ranganathan, C., Ganapathy S., 2002. Key dimensions of business-to-consumer web sites. .Information & Management. 39: 457. Sait, S. et al., 2004. E-commerce in Saudi Arabia: Adoption and perspectives. Australasian Journal of Information Systems 12(1): 54-74. Siala, H. et al., 2004. The impact of religious affiliation on trust in the context of electronic commerce. Interacting with computers. 16(1), pp 7-27 Strauss, A. and Corbin, J., 1990. Basics of qualitative research grounded theory procedures and techniques. USA: sage publications. US Department of Labor. 2008. US department of labor in the 21 century. [Online] Available at: dol.gov/asp/media/reports/chartbook/2006-06/ section1_txt.htm [Accessed 28 December 2008]. Vatanasakdakul, S. et al., 2004. What prevent B2B ecommece adoption in Developing countries? A socio-Cultural perspective. Proceedings of the 17th Bled eCommerce Conference on eGlobal, Slovenia, 2004 World Bank Group (2006), Saudi Arabia Data Profile, World Devlopment Indicators Database, Pittsburgh, PA. 18 IADIS International Conference e-Commerce 2009 DOES FIT B2B E-COMMERCE FOR AGRIBUSINESS? A FIRST APPROACH FOR TRUST EVALUATION ISSUES IN SPANISH AGRIFOOD SECTOR Fernández Mª Cristina, Research Student. PhD Candidate* De Felipe, Isabel. Prof.* Briz Julián, Prof.* Agricultural Economics Department. E.T.S.I. Agrónomos. Universidad Politécnica de Madrid. SPAIN Work in progress in the frame of the specific support action of the 6th European Framework Program named “E-Trust” (FOOD-CT-2006-043056). ABSTRACT The lack of information about trust issues in agricultural and food sector follow the authors to explore about this topic. A qualitative study is carried out in two phases in order to find out evidences for later hypothesis development about the performance of trust in the agricultural and food sector. This paper goes a step further in a previous one presented in ecommerce IADIS conference in 2008. Firstly this paper shows and initial reflection of the barriers related with trust in the adoption of B2B strategies and technological tools in agricultural and food in Spain. In order to explore this topic nine qualitative interviews where conducted. The result of this first stage is a list of qualitative results related with concerns and barriers in the e-business agricultural sector. The second phase consists on nineteen qualitative interviews using the Analytical Hierarchy Process (AHP) to evaluate a developed theoretical topology of trust. Finally the results are expressed and resume in the form of hypothesis to be study in forthcoming phases of the study. The results will be useful for the approach and implementation of trust in different supply chains in different agrifood sectors. KEYWORDS Trust, AHP, Agribusiness B2B. 1. INTRODUCTION The Agricultural, Food and Beverage sector which collect the activities related with production and transformation from raw products to human consumption, is one the most important industrial branches in Europe and Spain. In 2004, the EU-25 food and beverage industry as a whole had a turnover of 815 billion euros, transforming over 70% of EU’s agricultural raw materials and employing about 3.9 million people, of whom the majority works in SMEs. France, Germany, Italy, the UK and Spain are the largest EU-25 producers with more than 70% of total EU turnover. However, this key sector presents a low rate of Internet and Communication Technologies (ICT) adoption than other sectors and less development of Business-toBusiness strategies. (European Commission, 2006-2007). The low adoption rate adoption of ICT although the theoretical benefits are stated in terms of promotion of information flow, transparency market and prices, reduction or elimination of transaction costs and the increase in online cooperative (Ferentinos, K. et al. 2006) leads to a reflection of e-commerce adoption in agricultural and food sector. Many characteristics of food products may only be analyzed after use (experience characteristics); others even cannot be examined at all (credence characteristics). Furthermore, the lack of physical inspection of the product and contact between transaction partners make e-commerce too anonymous for agrifood sector transactions also may cause a lack of trust, which could be responsible of the low adoption in the sector. (Fritz et al. 2007) 19 ISBN: 978-972-8924-89-8 © 2009 IADIS In terms of the communication technology adoption there are lots of debates around the application of Information Technologies in the Supply Chain Management but a very few literature survey article that deals ICT with Supply Chain Management (Gunasekaran, Ngai, 2004) with Information Technologies aimed to ebusiness and focused in agricultural and food sector. This report shows a preliminary approach in a qualitative way to explore how the communication, networks and B2B relationships works in the agricultural and food sector. 2. BACKGROUND OF THE QUALITATIVE STUDY 2.1 Trust and E-business Activities in the Agrifood Sector. Theoretical Framework The Electronic Commerce environment is an environment with risk, simply because commerce in general involves risk (Tan and Thoen, 2001), the agrifood sector deals with products and activities where a large amount of information asymmetry exists between transaction partners (Fritz M. 2007) and the adoption of EBusiness transaction support by businesses is low, in particular by small and medium sized enterprises (SMEs) in food networks (European Commission, 2006-2007). Tan and Thoen (2001) argue in their “Generic Trust Model” that the agent’s trust in a transaction with another party is the combination of the trust in the other party and the trust in the mechanism for the successful performance of the transaction. Fritz M, Hausen. T. and Cannavari M. advanced going further towards a “Trust Model for Electronic Commerce in the agrifood sector” based on Tan and Thoen. The level of trust in a transaction is depending of elements which generate trust such as other party and control mechanism. Thus, the level of trust in a transaction is depending of the situation where the transaction is being carried out. The potential gain and risk of transaction the risk attitude of the individual who perform the transaction determine the situation where the level of trust is set. determined by scenario determined by scenario; dynamic interrelation with trust generation determined by individual Figure 1. Trust Model for Electronic Commerce in the Agrifood sector based on Tan and Thoen (2001) developed by Fritz, Hausen and Cannavari (2007) 2.2 Barriers to the B2B Adoption Barriers to the adoption of in B2B environments are mentioned in different reports and studies. The European Business report 2006-07, shows the following barriers not focused on agricultural and food sector: company too small, technology too expensive and complicated, incompatibility systems, legal issues and lack of reliable IT suppliers. Other Spanish National studies carried out by AECE (AECE, 2001) and the ECommerce Observatory of Madrid Chamber of Commerce (Cámara de Madrid, 2003) added to the previous some more specific to Agricultural and Food sector like: lack of standardized product and knowledge (specific human resources formation), company culture (culture endurance and narrow minded), some 20 IADIS International Conference e-Commerce 2009 companies prefer to continue with old strategies, the ratio cost/benefit does not justify the investment. Other barriers related with an e-business environment are [E-Thematic, 2003] deal with the organization (uncertainty about business models, the agrifood sector should deal with specific business models according to the food sector characteristics), the company operations (operational changing procedures, lack of technological and human skills and fragmentation in the software market) and legal and jurisdictional issues (difficulty to understand how the law is applied to e-commerce because in some areas such as digital signatures, tax laws, customs tariffs, the confusing in disputing resolutions and infrastructure and security issues. External related factor like trust, security, successful relationship, internet affordable access and customer acceptance are considered external critical B2B success factors (Riyad. E, Myfanwy T, Abdel. M, 2002). The role of trust in the agricultural and food sector is not mentioned explicitly in previous studies but it is suspected that the lack of trust may be considered a threat for developing e-business strategies and it is considered that the lack of trust it is not enough explored in agrifood sector. This the start point in the research. 2.3 The Trust Typology It seems to be a fact that a barrier for the electronic transactions of the e-business in the agricultural sector is the difficulty to test and to examine the product. There is also a perceived reluctance in the sector to perform transactions and relationships via e-commerce. In the frame of “E-trust” project a theoretical typology developed by Hofstede G. J et al (2007) provides a basis to test the dimensions of trust with acultural point of view. The main characteristic of the trust typology is the wide view of “trust” (including “control” o it), it is focused on food quality and safety attributes, it is focus on early stages of relationship, it is taken the perspective of the buyer, it is based on state-of-the-art, its structure it is inspire for AHP method use, it is flexible across sectors and elements are included to light potential sources of cross-cultural differences. The 1st hierarchy level consists of the main goal of the typology, trust from the perspective of a buyer who is in the early stage of a new purchase relationship. The 2nd level consists of the objects of trust (product, seller, market environment). This level is expected to be cultural sensitive; the importance of control institutions is expected to be strongly associated with product characteristics, as well as culture. The 3rd and 4th level contain the dimensions of trust. Objective Objects of Buyer's trust in trust 1. Product transaction Dimensions of the objects of trust 1.1 Reputation 1.2 Specification 1.3 Inspection 1.4 Certification 1.5 Price / performance ratio 2. Seller 2.1 Capability 2.2 Relationship 2.2.A Relationship between individuals 2.2.B Relationship between companies 2.3 Reliability 2.3.An adequate communication 2.3.B Deliveries 2.3.C Financial situation 2.4 Reputation 3. Market 3.1 Control institutions environment 3.2 Informal institutions 3.3 Legal institutions 3.4 Reputation Table 1. Basic Structure of the Trust Typology (Source: Oosterkamp and Hofstede, 2007) 21 ISBN: 978-972-8924-89-8 © 2009 IADIS 2.4 AHP The Analytic Hierarchic Process (AHP) is a widely used method to solve complex decision problems, for scientific as well as for business applications The AHP was introduced by Saaty (1980) in order to structure and solve complex decision problems. However, it may also be used to derive priorities for hierarchical elements which cannot be deducted from evaluator’s experience. The application of the AHP is more or less independent from branches and business fields and is used continuously for decision problems in agriculture. (Amedeser, Haas, Meixner 2008). The AHP promotes to divide the problem into smaller elements which are incorporated in a predefine structure such a in a hierarchy tree (Amedeser et al 2008). The elements of the hierarchy tree are based of the structure of the typology. The objects of trust and the dimensions of the typology are used to be weighted by “pairwaise comparisons”. The comparisons of element according to an established scale allow to join the qualitative information with quantitative one. The final result is a numerical value expressed in percentage which presents the most weighted option of the hierarchy. 3. METHODOLOGY The methodology to study B2B relationships in agrifood sector in Spain followed two separated ways. On the one hand the scarce study about the performance of trust in Spanish agribusiness followed to make an exploratory study to know about the concern and worries about the implementation of business to business strategies. In the other hand questionnaire using the AHP were carried in order to check which were the most important attributes and values developed previously in the typology and link them with the exploratory qualitative interviews. 3.1 Qualitative Interviews In order to develop a qualitative approach towards trust in e-commerce in agrifood sector framed in the “Trust Model for Electronic Commerce in the Agrifood sector”, ten interviews were conducted with representatives and professional of ICT technologies of the sector in Spain. The purpose is to know how ebusiness and trust is performed to have a broad overview on the main problems faced in adopting ecommerce, interviewed were convenience chosen. The questionnaire consists on a short list of twelve open questions (see table 3). The main issues asked to the interviewees where related with the need, opportunities, the preconditions for e-business adoption and the suitability of e-commerce for the agrifood sector. The types of representatives interviewed are listed in the table 2. The interviews were transferred to an extended summary. The results obtained by the interviews make up a previous and qualitative scenario for later quantitative studies. Table 2. List of Conducted Interviews SPAIN Sector Meat Convenience Products Stage Manager of Sales Sector Fruit cooperative sociedy Stage E-business manager and cooperative member Food Consortium Responsible for trade marketing Grain Snack Transformer Quality Manager Olive Oil Brand Marketing Manager Academic University Business teacher Expert (focus: computer consulting) Strategic ICT Consulting Manager of Consulting Company Computer consulting 22 IADIS International Conference e-Commerce 2009 Table 3. Questionnaire for the Qualitative Interviews 1. Do you use E-commerce in your business transactions? If YES, 1.1 When did you start? a) Less than 1 year ago b) Between 2 and 5 years c) More than 5 year 1.2 Which proportion of your business account e-commerce? (%) 1.3 Do you expect to increase e-commerce in your company? 1.4 Do you have a web page? 1.5 Others. If NOT, Which are the main reasons? 1.6 No interested in new technologies. 1.7 Difficult to understand 1.8 Do not suit my product- service requirements. 1.9 Difficult to operate. 1.10 Very expensive 1.11 Others 2. Please, identify the main elements which give you: Trust in e-business (What elements do you aim to trust in e-business? (I.e. Brand reputation, certification, legal control, on time deliveries). Distrust in e-business. (What elements do you aim to do not trust in e-business?)(I.e. Brand reputation, certification, legal control, on time deliveries). 3. What measures/technical elements give/would (if e-processes are not used) give you trust in your e-processes? (Web cameras, microphones, videoconference…etc.). 4. What elements/procedures do you miss/need in your e-processes which would give your more trust in them? 5. How do you create/perceive a “trusted market environment” for selling/buying your products) (Keep in mind typology 3. market environment: control institutions, informal institutions, legal institutions, reputation) 6 Describe which are the main factors to select your suppliers/customers to trust in them (Quality, price, location, confidence, reputation…) 7 In your opinion which are the main obstacles to use Internet and trust in the transactions (High cost and tariffs, lack of personnel expertise, lack of tangible, benefits…) 8 .In what business activities do you use e-commerce? In what activities do you think that e-transactions does not work related to trust? (Request of goods and services, to carry out payments, electronic reception of good and services…) 9 What barriers do you find for buying by e-commerce related with trust? (Difficulties for some products and services, small number of suppliers, cost of delivery, logistic problems, payment and contracts uncertainty…) 8 What advantages do you find when buying-selling by e-commerce? (Cost saving, speed of the processes, simplification of the tasks, greater number of suppliers…) 9 How do you sell or develop activities by e-commerce? What do you show your partners to trust in you? (Through the web page and/or e-market place, description of product and services, certification, stamps) 10 Which are the activities developed by firms selling through e-commerce (information of products, services and prices, get orders and payments, deliver goods and services…) 11 What problems do you find in firms selling through e-commerce? (Lack of potential clients, payment uncertainty, contracts uncertainty, logistic problem, maintenance cost of the system, high priority of traditional channel…) 12 What are the advantages for selling through e-commerce? (Cost reduction, get new clients, market geographic expansion, better quality of the service, higher speed of the process, simplification of tasks, to avoid to yield quota market to companies already operating in e-commerce…) 3.2 AHP Assessments The Trust typology was tested by 19 interviews assessment using a software tool. Although it exists specific software designed for multicriteria choice decision it was decided to developed the AHP questionnaire by the excel tool. Excel is very common software, available in most of the computers and it allows to the respondents to answer easily by email. In this excel file the different object of trust typology were evaluated, pondered and weighted by the chosen respondents (see table 4) by using the excel tool. The excel tool provides interaction to the interviewed person because the pondering is very intuitive moving and locating the cursor in the right way. Interviewees also provided some extra information during the evaluation by faceto-face or telephone conversation. The excel tool provides interaction to the interviewed person because the pondering is very intuitive moving and locating the cursor in the right way.The scale for the evaluation process is a standardized AHP scale confirming Saaty (1980) with verbal description of the data points (from 1 = equal importance to 9 = absolute dominating and the associated reciprocal values for reversal evaluations). Figure 2. Evaluation of Trust Typology (Via Excel) 23 ISBN: 978-972-8924-89-8 © 2009 IADIS Table 4. Conducted AHP Assessment Type of Company Company Level in size (number value chain of employees) Function of the respondent GrowerWholesaler 2 Wholesaler Transformer More than 500 NationalICT manager international Transformer Retailer 2 local Owner/manager Growerbuyer 9 localregional national national Manager Quality Manager Transformer National Technical Director GrowersQuartering Transformer 5500 Transformer -- Wholesaler 20 Transformer- 87 wholesaler Olive oil Bottler GrowerBottler 2nd grade cooperative COAG UPA More than 500 National Marketing Manager International 20 National Sales Manager +cooperatives international National Sector Managers Fruit & vegetable national S e c t o r Meat Grain S e c t o r Type of Company Company Level in size (number value chain of employees) Function of the respondent Dairy Product Transformer * Cooperative More than 1000 national Directive and Consulter 29 growers +technical staff 2 international Cooperative member national Manager Around 20 national Manager Around 100 (depending season) 5 nationalPurchase international Manager Family enterpriseGrower Wholesaler regional Veterinary Regional Key Account international Manager national Quality Manager National Director 4. RESULTS DISCUSSION 4.1 First Phase Regarding to the requirements for not adopting the ICT tools in agricultural and food B2B sector some explanations appeared. B2B is not aimed for all products in food sector. It suits for those products not perishable with high added value and with low distribution cost in relation with the product price (i.e., wine, high quality olive oil, traditional products…). In order to supply the raw inputs the e-relations becomes more complicated in terms of trust to assure the organoleptical attributes of the raw products (quality, freshness, taste…). The degree of differentiation of some agrifood products makes the e-transactions very difficult, and when the business relationship concerns with not standardized products e-commerce does not work easily. Other important characteristic that influences the adoption of B2B is the intrinsic and social characteristics of agricultural and food sector. The traditional way of transacting (knowledge by acquaintance and physical approaches) seems to be the most guaranteed and proofed one. Moreover, if there is not a common intention and a strict request (by clients) to accesses to a new way of transacting, no concrete advantages seem to come from accessing e-commerce. Operators underlined also the need for “signals” in ecommerce, in order to be assured (institutional, legal, financial signals). Information and communication are key factors which can help the development of e-commerce, but reputation of business operators and their position in traditional markets seem to be still considered very relevant in the decision to access emarketplaces and e-environments. However, under the “umbrella” of e-business a lot of administrative transactions are carried out specially exchange of information, search of new suppliers and data exchange. Within the interviews some preconditions for e-business adoption were able to be found out. One of the preconditions is getting standardized products with enough quality attributes. The less suitable products to ebusiness are those food products less processed and with less long life time. 24 IADIS International Conference e-Commerce 2009 The interviewed persons have a special concern about trust issues. They showed themselves worried about how and where complaining and the necessity of physical documents to complain. Moreover, there are worries about the ignorance of the players who take part in the business process, the possibility of phishing and the possibility of spying or intromission in key company data. 4.2 Second Phase The main conclusion found it after the nineteen interviews carried out is that major weighted objects of trust depend on the place that the company is located throughout the supply chain and it does not depend on the sector. Therefore the most important object of trust typology is the product, considering in most of the cases raw material, for transformer companies which import raw materials and for growers is the reputation, reliability and the ability to get the money of the purchase at time. Concerning the three levels of the assessment in Spain the results can be synthesized as follows: At first level of the assessments the major pondered objet were the product, in second place the “company” and in the third last place the “market environment”. At second level of the assessments the main mentioned object of trust where: “reputation” and “price-performance” for grain sector, “inspection” and “price performance” in fruit sector, “inspection”, “reputation” and “specification” in meat sector and finally “price performance” and “specification” in the olive oil sector. At third level the major pondered trust objects are the “relations between individuals” and the “deliveries”: Regarding the evaluation between public institutions and private institutions the major weight belongs to the public ones. However, some of interviews pointed out that for specific product some private ones work much better. 5. CONCLUSIONS AND FURTHER RESEARCH Interviews show that e-commerce needs some more time to be culturally accepted in agricultural sector. The interference with traditional ways of doing business is seen as a real problem, together with the lack of the perception of virtual environment as a business tool. Institutional guarantees are seen as important, but reputation seems to play a key role in this environment, together with marketing and business players’ decisions. Technological aspects need to be improved in order to better satisfy each products’ needs. Regarding to the suitable agricultural products the interviews show that entrepreneurs are reluctant in adopting new technologies, especially in fresh products because it is a big challenge to make perceptible the freshness quality attributes of the products to the potential customers. E-commerce is more suitable for those products with high added value and with possibility to be easily stored. Regarding the virtual environment, it is mentioned a special worry about the publicity of the personal data, business partners, etc and to be vulnerable to “electronic mistakes”. A legal environment where the rules of the game where establish is very important. It is very important as well to have “evidence” and proves just in case to complain. Physical contact still remains as a very important issue. An important way to make business is starting from a traditional way and continue with this partner in an “e-business way”. The dimension of the typology “Product” is the main weighted object of trust for the interviewed using the AHP assessment excel file. It is especially important for those companies that use products as raw material. There are no very high differences between the different sectors, the difference appears between those companies which are considered as transformers and demand a high quality product those considered as growers which demand a correct price at an agreed time. However, the company plays an important role of adding value in the same sold product. Finally, in general terms the market regulations developed by public bodies are considering more trusted institutions than private institutions. This research present the progress in the field of the application of information and communication technologies and the performance of trust in e-commerce in the agricultural and food sector. 25 ISBN: 978-972-8924-89-8 © 2009 IADIS For future research several hypotheses should be proposed in order to develop a further quantitative study focused in a specific sector. H1: B2B e-commerce is aimed for agricultural standardized added value food products. H2: Agricultural and food sector B2B relationships starts with a traditional way of making business where physical contact is still important. H3: “Product” is the core tangible term which provides trust in B2B agrifood sector transactions. H4 : “ Reputation” in the core intangible term which provide trust in B2B agrifood sector transactions. The hypothesis represents a basis to go deeper in the task of indicators that provide trust in the transactions along the supply chain and the attributes of the products that can be traded by e-commerce. REFERENCES AECE, 2001. Resumen sobre Comercio Electrónico B2B.Observatorio del Comercio electrónico. España Cámara de Madrid. Comunidad de Madrid. España. Amedeser, C. et al. 2008. Measurement of the importance of trust elements in B2B AgriFood Chains, an Application of the Analytic Hierarchic Process. Journal on Chain and Network Science; 8(2) pp 153-160 Briz,J. Schieffer; G. Fritz. 2008 Report on use of B2B trust elements in e-commerce Unpublished report, EU FP6project: “e-trust”, contract number FP6-CT-2006-043056 Cámara de madrid, 2003. Comercio Electrónico en la Industria Alimentaria de Madrid.Observatorio del Comercio electrónico. España.Cámara de Madrid. Comunidad de Madrid. España. European Commission, 2007 The European e-Business Report Available on http://ebusinesswatch.org/key_reports/synthesis_reports.htm ISBN 92-79-02038-2, pp20-34 European Commission, 2006 ICT and e-Business in the Food and Beverages Industry. Available on http://ebusinesswatch.org/links/sectors/food.htm e-business W@tch, European Commission; Databank S.p.ASector report Nº 1/2006 E-Thematic, 2003.State-of-art report on e-fulfillment. Thematic Network Ferentinos, K. et al. 2006. Internet Use in Agriculture, Remote Service and Maintenance: E-Commerce, E-Business, EConsulting, E-Support. Handbook of Agricultural Engineering Chapter 7 pp 453-464. Fritz, M. Hausen, T. Cannavari M. 2007. Trust and e-commerce in the agrifood industry: Configuration of a trust environment for e-commerce activities. In: Theuvsen L et al. (eds) Quality Management in Food Chain. Wageningen Academic Publishers, Wageningen. Gunasekaran A, Ngai E.W.T (2004). Information System in Supply Chain Integration and Management. European Journal of Operational Research 159 269-295. Ho W. Integrated analytic hierarchy process and its Applications – A literature review. European Journal of Operational Research 186 (2008) 211–228 Oosterkamp, E. and Hofstede, G. J. 2007. Report on B2B trust elements and their typology. Unpublished report, EU FP6project: “e-trust”, contract number FP6-CT-2006-043056. Riyad E., Myfanwy T. and Abdel M. 2002. Across Industry Review of B2B Review of Critical Succes Factors. Electronic Networking Applications and Policy Volume 12 . Number 2 . 2002 . pp. 110±123 ISSN 1066-2243 Saaty, T. L, 1980. The analytic hierarchy process: planning, priority setting, resource allocation. New York, NY, McGraw-Hill. Tan, Y.-H. and Thoen, W (2001). Toward a Generic Model of Trust for Electronic commerce. International Journal of electronic Commerce 5 (2): 61-74 26 IADIS International Conference e-Commerce 2009 STRATEGIC ALIGNMENT AS A WAY OF ADDRESSING THE BARRIERS TO E-BUSINESS ADOPTION Eduardo Escofet Department of Information Technology, University of Holguín Ave. XX Aniversario s/n, Holguín, Cuba María José Rodríguez, José Luis Garrido Department of Computer Languages and Systems, University of Granada ETSIIT, c/ Pdte. Saucedo Aranda s/n, 18071 Granada, Spain Lawrence Chung Department of Computer Science, University of Texas at Dallas Richardson, Texas 75083, USA ABSTRACT IT managers are key human resources in E-Business, as strategic thinking blocks in E-Business adoption at IT levels. Despite the critical role IT managers play, there has been inadequate support for aligning E-Business with IT. Although some tools and techniques have been proposed in literature to find and evaluate Business and IT Strategic alignment, they suffer from lack of objectivity and integration. In this paper, we propose a four-step interview-based method that permits us to align E-Business with IT strategies and to obtain investment priorities per software development process and area, towards improvements on service quality. This method incorporates the Delphi Method, the Goal Modeling formalism, Venkatraman´s Strategic Alignment Model and the QFD technique to avoid inconsistencies and to increase confidence in compiled data and modeling results. KEYWORDS E-Business / IT alignment, adoption barriers. 1. INTRODUCTION While adopting E-Business technologies, we face a clearly identifiable barrier: lack of strategic thinking. This barrier is present at all business levels, in particular at the IT level. Aligning E-Business and IT strategies is a way of addressing this barrier and also to positively impact other barriers such as troubles with developing an E-Business roadmap and the lack of management commitment. The E-Business/IT strategic alignment is an essential step for architectural, structural and new technological adoption decisions. Several problems emerge in this process: 1. Interviews to business and IT managers have a high level of incoherence, due to poor skills on new technologies and raw knowledge of on-line business models. 2. Applied questionnaires contribute with a high level of subjectivity to the alignment process. 3. IT investment priorities are not clear; this is a key element for the success of the E-Business adoption. We can find several studies about the E-Business/business/IT alignment subject in [Al-Hakim 2009], [Baina 2008], [Becker 2008], [Henderson 1999], [Pierre 2008], but they show lack of integration, obviate priorities on investment and specific features of E-Business model, or usually forget the human-subjective aspect of each alignment process. We propose a four-step, interview-based method, starting with the preparation of the interview questionnaires, and then applying the Delphi method [Linstone 2002]. This is a method used to reduce incoherence in the interviewing process, to improve consolidation results and to reach consensus. 27 ISBN: 978-972-8924-89-8 © 2009 IADIS Next, we apply the goal modeling formalism and the Strategic Alignment Model (SAM) relationship identification by comparison to reduce subjectivity on the compiled information and to guarantee strategic EBusiness/IT alignment. Last, we relate identified goals with processes and process areas (for example, identified from a CMMI application) using Quality Function Deployment (QFD) to align strategic goals and IT processes in an organization [Becker 2008], and to determine investment priorities in IT processes and process areas. To clear concepts and steps we use an example through our method unfolding. 2. UNFOLDING THE METHOD The proposal of this method, oriented to facilitate the alignment between E-Business and IT strategies, is illustrated using an example, the fictitious case of ebooks2go.ws portal, similar to other portals like freebookspot.org or knowfree.net. It receives thousands of hits everyday and permits to search for e-books and e-zines from Internet content providers such as megaupload.com or rapidshare.com. Its web site offers well known typical basic services for this kind of business: simple searching and scrolling features, registration and authentication forms, content descriptors, external content links, and front page pictures when available. Its main business is based on advertisement links, usually from Google Ads or sites alike, with all of these attributes organized across a plain web interface. 2.1 Step 1 – Interviewing and Consolidating Our proposal is interview-based and model-oriented, so the first thing to ensure is the accuracy of the questionnaires and the reliability of the resulting data. We would reach this previous achievement if experts in questionnaires and interviews focus their attention on objectives through the whole interviewing process. It is of paramount importance to count on this in the following refinement process for strategic goal modeling. The interviewing process should involve business managers, IT managers, stakeholders and end users, in order to have a whole understanding of the needs and capacities. A deeper discussion on this subject is beyond the scope of this paper, but we encourage people interested in gathering methods to study other materials about questionnaire preparation and interview planning [Esposito 2002] [Frary 1996]. It is important to notice that we need other methods to reduce incoherence and to reach consensus. We recommend a study on elicitation techniques adequacy [Carrizo 2008] to better grip effectiveness of these methods. While applying Rappa´s [Rappa 2004] [Afuah 2003] and Weill & Vitale´s [Weill 2001] E-Business model taxonomies, we have found that this E-Business model should be classified into the Advertising and the Intermediary groups, thus clarifying some of its features and attributes. It is important to identify EBusiness models in order to increase the accuracy of the proposed method and to consider the specifics of each model in the processes of questionnaires preparation, interviewing and information consolidation. A thorough analysis of the answers to the questionnaires should be done in order to attain a better understanding of real E-Business needs, end users and stakeholders preoccupations, business managers opportunities outlook and IT managers identified IT capacities.. 2.2 Step 2 – Applying the Delphi Method Applying questionnaires and interviews usually produces heterogeneous and inexact data that should de compared and verified to increase information accuracy. The Delphi method is intended to reduce incoherence, increment objectivity on subjects and data stability, and help attain consensus. Before applying this method, it is essential to determine the objectives to obtain, to ensure the questions in the questionnaires conform to the objectives, to monitor the way surveys and interviews emerge. Here, the whole process has be to be centered around the strategies of E-Business and IT. We suggest this summarized steps be applied with the Delphi method in an interview-based process: Conformation of a Delphi team to supervise the entire project, selection and clustering of the informants (usually informants are experts in the research area), first round of Delphi´s questionnaires application, test questionnaires for proper terminology, transmission of the first round questionnaires to the jury members, analysis of the first round answers, preparation of the second 28 IADIS International Conference e-Commerce 2009 round of questionnaires, transmission of the second round of questionnaires to jury members, analysis of the second round answers (iterate through the three previous steps, including this one, to increase results stability), and preparation of a conclusion report by the analysis team. A deeper discussion on the Delphi method, especially on its associated statistical techniques, is beyond the scope of this paper, but we encourage the reader to take a look at some theoretical and practical studies [Linstone 2002], [Hsu 2007], [Yousuf 2007], [Rayens 2000], [Paul 2008] used in our research. It is important to reduce interviewing noise in consideration of E-Business model features and attributes. As a result of this step, we detected and eliminated most of the incoherence, inferred new requirements and relationships between requirements, and obtained a filtered subset of features, in particular, useful requirements for subsequent modeling. 2.3 Step 3 – Modeling and Aligning In this step, we use visually-oritend goal modeling, to reduce subjectivity in the compiled information and to enhance strategic E-Business and IT alignment [Baina 2008], [Bleistein 2004], [Lamsweerde 2001]. After that, it is essential to establish the relationship between the goal model and the relationships in Venkatraman´s Strategic Alignment Model by comparison. Four perspectives can be identified in SAM as follows: 1. Strategy Execution: This perspective corresponds to the classical, hierarchical view of strategic management. It considers the business strategy as the driver of both organizational design choices and the logic of the IT infrastructure. Top Management formulates the strategy; IT Management is only considered as strategy implementer. 2. Technology Potential: This perspective also views the business strategy as the driver. However it involves the formulation of an IT strategy to support the chosen business strategy and the corresponding specification of the required IT infrastructure and processes. The top management should provide the technology vision to articulate the logic and choices pertaining to IT strategy that would best support the chosen business strategy. The role of the IT manager should be that of the technology architect. He designs and implements efficiently and effectively the required IT infrastructure that is consistent with the external component of IT strategy. 3. Competitive Potential: This alignment perspective is concerned with the exploitation of emerging IT capabilities to: impact new products and services, influence the key attributes of strategy, as well as develop new forms of relationships. Unlike the two previous perspectives, which considered business strategy as given, this perspective allows the modification of business strategy via emerging IT capabilities. 4. Service Level: This alignment perspective focuses on how to build world class IT organization within an organization. In this perspective, the role of business strategy is indirect. This perspective is often viewed as being necessary, but not being sufficient, to ensure the effective use of IT resources and to be responsive to the growing and fast changing demands of the end-user population. Note that businesses with a large and dominant presence of the first perspective also have several barriers to E-Business adoption. Businesses with a dominant presence of the fourth perspective are usually technology-driven, hardly meeting user needs. A reasonable combination of perspectives with a large presence of the second and third perspectives is greatly recommended; otherwise some alignment and adoption problems exist and must be addressed. In this case, to reduce subjectivity on the interview/Delphi process, to model goals and to align EBusiness and IT strategies, we partially used the proposal of Baina [Baina 2008]. It permits to establish an adequate formalization of both strategies alignment with the difference that we start from consistent information resulting from the application of data elicitation techniques. It permits us more accuracy in modeling and to determine actual perspectives in the strategic alignment model [Henderson 1999]. Increasing accuracy in modeling E-Business/IT strategies is not easy if managers and experts don´t follw standards, so we use Weill & Vitale´s and Rappa´s taxonomies to classify the E-Business model of the organization; beginning with this classification it is easier to identify fundamental features and attributes of E-Business models. Goal modeling is an important part of the requirements engineering, we encourage to dig notation and formalisms foundations in several studies about goal modeling and goal-oriented requirements engineering (GORE) [Pierre 2008] [Lamsweerde 2001] [Anwer 2006] [Lamsweerde 2004] [Chung 2000] used in our 29 ISBN: 978-972-8924-89-8 © 2009 IADIS research. The following two subsections address the specific activities to formalize and model compiled interview results looking for the desired strategic alignment. 2.3.1 Goal and Business Modeling In this step, we started from the compiled interview results after the Delphi method application and extrapolated E-Business and IT strategies to goal modeling. As a result of the compiled information obtained from steps 2.1 and 2.2, we established that the main business goal is to increase profits through the growth of its user base (cloud in Figure 1). Because ebooks2go.ws revenue comes from the advertisement it shows, the main concerns according to interviewing results are increasing on-line ad number is related to clicks on ads by visiting users, the time of those users spent on searching for e-books and e-zines, and the site registered user base increment, as shown at the second level in Figure 1. An important E-Business gauge is the report from content provider businesses, which includes important data about the origin of a downloading, service level agreement (SLA) satisfaction and current content indexing levels of the site. After these steps, we have our goal models, as in Figure 1 with E-Business strategy and in Figure 2 with IT strategy. These models show relationships, dependencies and refinements of E-Business and IT objectives, using one of the prominent goal modeling methodologies [Lamsweerde 2001, 2004]. Increase profits + Increase pay-per-click ads + + + Increase registered users Info update customization Increase clicks on ads + + Increase permanence time Increase visits Increase quality and Amount of indexed materials Get higher site visibility Fig. 1. E-Business Strategy In Figure 2 we can see some of the identified IT goals, most of them with a cause-effect relation with EBusiness goals, and also important technology-related goals such as: Automatic Customer Service and Efficient and effective indexing mechanism, both at the second level in Figure 2, which assure the appropriateness of the IT infrastructure to properly achieve high level strategic goals. 30 IADIS International Conference e-Commerce 2009 E-Business support + Automatic Customer Service + Efficient and effective indexing mechanism Search engine optimization Secure authentication mechanism Bulletin subscription mechanism Content provider booking Powering IT capacities Fig. 2. IT Strategy 2.3.2 Aligning by Comparison After obtaining the goal models, we establish a comparison between the identifiable elements in order to build the strategic alignment model (SAM). This stage is largely based on the modeler´s skills and expertise, using the reviewed technical literature and our own experience, and some best practices recommended. This stage establishes semantic relations between all pairs of elements (E-Business element, IT element), while classifying relationships according to Venkatraman´s perspectives taxonomy and determining cause-effect relations where you can easily find orientation by examining element precedence in the compiled results of the interviewing process. Table 1. Comparison of the E-Business/IT Strategies IT Element Orientation Business Element a E-Business support ← Increase profits b Automatic customer service → Increase registered users c Efficient and effective indexing mechanism → Increase permanence time d Efficient and effective indexing mechanism → Increase visits At this moment the Strategic Alignment Model can be implemented. Three types of alignments are identified according to Venkatraman´s perspectives taxonomy: a) The main soft-goal of Increase profits will demand constant support from IT managers, developers and technologies. b) The creation of an Automatic customer service defines a publish/subscribe customer relationship management (CRM) style, incrementing the possibility of the creation of a faithful users base. c, d) An efficient and effective indexing mechanism permeates all systems but specifically help to increase user permanence time and visits, improving E-Business services without big E-Business strategy changes. From Table 1, we derive a Henderson & Venkatraman´s model, as shown in figure 3. 31 ISBN: 978-972-8924-89-8 © 2009 IADIS a b IT Strategy Business Strategy c, d Business Infraestructure IT Infraestructure Fig. 3. Strategic Alignment Model 2.4 Step 4 – Quality Function Deployment From the previous step, we obtained the strategic E-Business and IT alignment. To further increase the quality of this process and results, we apply QFD. This is a tool to implement quality improvement processes in heterogeneous production and services environments. We used it to establish priorities in investment decisions. We take the House of Quality matrix (HoQ) to relate E-Business goals with the necessary IT elements. Then an initial weight is assigned to each E-Business and IT element, where weights are obtained from the compiled information from informants, in steps 2.1 and 2.2 [Becker 2008] [Jayaswal 2006] [Denney 2005]. The same process is repeated to assign weights and initial priorities to HoQ, as well as relation levels between E-Business elements and IT elements, usually between 0 and 9. The first HoQ is in Figure 4. To adjust priorities we calculate new adjusted priorities using the formula: Adjusted Priority = E-Business Goal Initial Priority * IT Element Initial Priority * ∑ Perceived Weights This is a columnar calculus and, for every column, we obtain real investment priority for IT elements and actions in order to better fulfill E-Business goals and increase return on investment (ROI). 35 50 5 9 3 4,5 1 3 8 3 2572,5 9 5025 E-Business Goals Initial priority Increase registered users Increase permanence time Increase visits Adjusted priority Efficient and effective Indexing mechanism Automatic Customer Service Initial priority IT Elements(actions) Fig 4. House of Quality The same process is applied in a second HoQ to relate tactical E-Business goals with problem causes, and in a third HoQ to relate tactical E-Business goals and the IT process areas attained from a level 2 Capability Maturity Model Integration (CMMI) process. 32 IADIS International Conference e-Commerce 2009 As a result of these calculations and adjustments, we obtained important numbers, for example, in Figure 4, in the first column of numbers: 5, 4.5 and 8 are the weights indicating initial priorities of E-Business goals for business managers, end users and stakeholders (directly from the interviewing process); and in the first row of numbers: 35 and 50 are the initial priorities of IT actions or elements according to IT managers replies. Numbers in the center of the matrix: 9, 1, 3, 3, 3, 9 express the expected cause-effect relation between IT actions and E-Business goals as shown in Table 1. Finally, in the last row, we attained the results of the application of the preceding formula expressing the adjusted priorities for IT actions. From this step of our method ,some improvements related to prioritization are obtained: 1. Best accuracy in IT actions priorities. 2. Secure a better ROI according to main E-Business goals. 3. Prioritize some processes and IT process areas. 3. CONCLUSION In this paper, we described a four-step, interview-based method, using simple yet useful techniques and procedures, while deploying some best practices. This method is intended to help with aligning high level business concerns with IT strategic concerns, utilizing goal modeling refinement and comparison. The application of this method is intended to consolidate more accurate and useful information, through the strategic E-Business and IT alignment, and to reduce the interviewing process subjectivity, through goal modeling and SAM visual formalisms. Additionally, we obtained a sketch of prioritization for IT elements and actions in order to increase ROI and to fulfill E-Business goals and end user needs. The visual definition of the method as a whole using Unified Modeling Language (UML) activity diagrams has been already built, although not described in the paper due to space limitations. Future work includes assessing, and improving, the effectiveness of the interviews, in relation to other elicitation techniques, in particular goal-oriented techniques. Another line of future research concerns a more systematic application of the Delphi method. More studies for other contexts, as well as for widespread and complex examples, are also to be undertaken. ACKNOWLEDGEMENT This work has been done thanks to cooperation of the Ministry of Foreign Affairs of Spain (AECID) and project TIN2008-05995/TSI of the Ministry of Education and Science of Spain. REFERENCES Afuah, A. and Tucci, C.L., 2003. Internet Business Models and Strategies Second Edition. McGraw Hill, New York, NY. Al-Hakim, L. and Memmola, M. (eds.), 2009. Business Web Strategy: Design, Alignment, and Application. Information Science Reference, Hershey, PA. Anwer, S. and Ikram, N., 2006. Goal Oriented Requirement Engineering: A Critical Study of Techniques. XIII Asia Pacific Software Engineering Conference (APSEC'06). IEEE Computer Society, Bangalore, India, pp. 121-130. Baina, S. et al, 2008. Petit, M., Castiaux, A.: Strategic Business/IT Alignment using Goal Models. In: Johannesson, P. (ed.): Third International Workshop on Business/IT Alignment and Interoperability, BUSITAL'08, Vol. 336. CEURWS.org, Montpellier, France, pp. 31-43. Becker, A.L. et al, 2008. Strategic Alignment of Software Process Improvement Programs Using QFD. BIPI’08. ACM, Leipzig, Germany, pp. 9-13. Bleistein, S.J. et al, 2004. Strategy-Oriented Alignment in Requirements Engineering: Linking Business Strategy to Requirements of E-Business Systems using the SOARE Approach. Journal of Research and Practice in Information Technology, Vol. 36, No. 4, pp. 259-276. Carrizo, D. et al, 2008. Study of Elicitation Techniques Adequacy. 11th Workshop on Requirement Engineering, Barcelona, Spain, pp. 104-114. 33 ISBN: 978-972-8924-89-8 © 2009 IADIS Chung, L. et al, 2000. Non-functional requirements in Software Engineering. Kluwer Academic Publishers, Norwell, MA. Denney, R., 2005. Succeeding with Use Cases Working Smart to Deliver Quality. Addison Wesley Professional, Upper Saddle River, NJ. Esposito, J.L., 2002. A Framework Relating Questionnaire Design and Evaluation Processes to Sources of Measurement Error. International Conference on Questionnaire Development, Evaluation, and Testing Methods, Charleston, SC. Frary, R.B., 1996. A Brief Guide to Questionnaire Development. ERICAE Clearinghouse on Assessment and Evaluation, Washington, DC. Henderson, J.C., Venkatraman, N., 1999. Strategic Alignment: Leveraging Information Technology for Transforming Organizations. IBM Systems Journal, Vol. 38, No. 2, pp. 472-484. Hsu, C.-C. and Sandford, B.A., 2007. The Delphi Technique: Making Sense Of Consensus. Practical Assessment, Research & Evaluation, Vol. 12, No. 10, pp. 1-8. Jayaswal, B.K. and Patton, P.C., 2006. Design for Trustworthy Software: Tools, Techniques, and Methodology of Developing Robust Software. Prentice Hall, Upper Saddle River, NJ. Lamsweerde, A.v., 2001. Goal-Oriented Requirements Engineering: A Guided Tour. 5th IEEE International Symposium on Requirements Engineering. Toronto, Canada, pp. 249-263. Lamsweerde, A.v., 2004. Goal-Oriented Requirements Engineering: A Roundtrip from Research to Practice. 12th IEEE International Requirements Engineering Conference. Kyoto, Japan, pp. 4-7. Linstone, H.A., Turoff, M., 2002. The Delphi Method Techniques and Applications. Addison Wesley, Reading, MA. Paul, C.L., 2008. A Modified Delphi Approach to a New Card Sorting Methodology. Journal of Usability Studies, Vol. 4, No. 1, pp. 7-30. Pierre, H. et al, 2008. A Method to Support the Alignment of Business Models and Goal models. In: Johannesson, P. (ed.): Third International Workshop on Business/IT Alignment and Interoperability, BUSITAL'08, Vol. 336. Montpellier, France, pp. 120-134. Rappa, M.A., 2004. The utility business model and the future of computing services. IBM Systems Journal, Vol. 43, No. 1, pp. 32-42. Rayens, M.K. and Hahn, E.J., 2000: Building Consensus Using the Policy Delphi Method. Policy, Politics, & Nursing Practice, Vol. 1, No. 4, pp. 308-315. Weill, P. and Vitale, M.R., 2001: Place to Space: Migrating to Ebusiness Models. Harvard Business School Press, Boston, MA. Yousuf, M.I., 2007: Using Experts’ Opinions Through Delphi Technique. Practical Assessment, Research & Evaluation, Vol. 12, No. 4, pp. 1-8. 34 IADIS International Conference e-Commerce 2009 ECUSTOMS CASE STUDY: MECHANISMS BEHIND COOPERATION PLANNING María Laura Ponisio1, Pascal van Eck1, Lourens Riemens2 1 Department of Computer Science, University of Twente P.O. Box 217, 7500 AE Enschede, The Netherlands 2 Dutch Tax and Customs Administration Apeldoorn, The Netherlands ABSTRACT Members of existing e-commerce trading networks constantly assess their network to identify opportunities for increased co-operation and integration of e-commerce IT systems. Failing to identify the mechanisms involved in co-operation compromises correct investment decisions. In this paper, we use Systems Thinking as a reasoning model that helps decision makers to uncover such mechanisms. We use Systems Thinking to analyse a real-world case called eCustoms, an inter-organisational network of customs organisations. The resulting model explains the mechanism of planning cooperation in terms of a feedback loop that comprises political support, operational potential, and information flow. This mechanism also explains why it is important to select potential partners for closer co-operation as early as possible, the importance of willingness to participate, and the gain or loss of decision power that joining a network implies. KEYWORDS Systems Thinking, Systems Dynamics, Inter-organisational networks, management practices. 1. INTRODUCTION Business-to-business e-commerce is supported by inter-organisational networks. These networks consist of information technology (IT) systems that support co-operation between trading partners, for instance in longterm strategic supply partnerships. Therefore, in e-commerce, understanding how co-operation works is essential to make correct investment decisions. Members of an established trading network constantly assess their network to identify partners with whom to increase co-operation, for instance to further integrate their e-commerce systems. Part of the problem is to identify other members that may be interested in closer co-operation early in the network formation process. There are several challenges in planning this co-operation process (Ponisio, Sikkel, Riemens and van Eck, 2007, 2008): possibilities for opportunistic behaviour (Williamson, 1993) have to be mitigated, potential partners must be assessed considering multiple perspectives, attention has to be paid to distribution of power in the network, the IT development process has to be planned, and sustainable gains have to be measured. The importance of such challenges is illustrated for instance in the well-known case of Covisint (Gerst and Bunduchi, 2005, 2007), where sub-optimal attention for power issues resulted in near-total failure. However, current systems development theories in e-commerce insufficiently address these challenges. An undesired consequence is that in practice, co-operation planning is done ad-hoc. Decision makers, therefore, ask for techniques that help them to uncover the mechanisms that determine successful co-operation. Uncovering such mechanisms should help experts to understand the forces and tensions involved, which should improve planning co-operation. For instance, which network partners to choose for further integration, on which topic to co-operate, and which additional systems to integrate first with existing partners? This paper presents a study of eCustoms (Section 3): a real-world case of an established interorganisational network in which customs organisations of the European Union (EU) co-operate electronically to improve ensuring safety of the EU external borders, and to facilitate trade. Members states of the EU form 35 ISBN: 978-972-8924-89-8 © 2009 IADIS bilateral or multilateral relations within the boundaries of this network in which IT systems are integrated. All these relations together shape the eCustoms network. Each relation is constantly evolving, having their supporting IT systems (and their architectures) evolving too. In this paper, we use Systems Thinking (Section 2) as a reasoning tool to explain the mechanisms behind co-operation planning. In particular, we are interested in uncovering the mechanisms that influence the choice of partners for intensified co-operation. Systems Thinking is a holistic problem solving method in which system behaviour emerges from the interaction of system components. Recent research proved the potential of using the Systems Thinking approach to explain mechanisms behind IT outsourcing projects (van Eck and Ponisio, 2008). This paper extends our earlier studies of electronic customs (Ponisio, Sikkel, Riemens and van Eck, 2007, 2008; Ponisio, van Eck and Riemens, 2008). Being able to reason in terms of such dynamics helps project managers to plan co-operation in interorganisational networks. The contribution of this paper is a Systems Thinking model that enabled us to identify causal loops in the mechanisms that play a role in co-operation planning (Section 4). We show how applying Systems Thinking helps the e-commerce network stakeholders to explain the rationale behind planning co-operation in inter-organisational networks. A better understanding of the dynamics enables decision makers to avoid surprises and instead predict the consequences of the planning actions they take. For instance, the Systems Thinking approach uncovered the need to anticipate exit strategies by adding an exit clause to the contract. In the case study, our findings indicate the importance of willingness (supported by, e.g., matching goals and matching needs), the importance of entering (or leaving) the co-operation network early, and that joining a closely co-operating group has – in addition to the benefits of working as a group – the potential to, in practice, lose decision power (because participants need to account for others in their decisions). We explain this using a concrete example of partner selection in eCustoms. We validate our findings via interviews with experts and conclude by presenting implications for research and practice. 2. SYSTEMS THINKING: BACKGROUND Systems Thinking (Midgley, 2003) analyses a system by modelling the relations between the components of the system and studying the behaviour that is jointly created by the interactions between these components. This creates a holistic view of the problem that helps to identify the dynamics of the system that results from interactions between its components. These dynamics would remain hidden in an analytic approach in which system components are studied in isolation. In previous work (van Eck and Ponisio, 2008), we have applied Systems Thinking in the domain of IT outsourcing projects. In that work, we observed that project outcome is not determined at a single moment in time only, e.g., at the end of the project, but by a process consisting of multiple interactions with members of the customer organization during the entire project. We developed a causal model (Figure 1) that explains the dynamics of project outcome. Outsourcing creates a relation between two economically independent actors (the outsourcer and the insourcer). The fact that the two actors are economically independent creates forces and tensions between them that in turn influence delivery quality as depicted in the model (words in italics refer to variables in Figure 1). During the project, both employ information flows to coordinate their activities. This information exchange, however, is influenced by the trust that they have in each other, which in turn is influenced by the perception of the outsourcer of the extent to which the insourcer is capable of delivery. As both actors are economically independent, we have to assume that each is self-interested and not unwilling to act in a way that advances its own interest in a way that is detrimental to the other (possibilities for opportunistic behaviour). The dynamic model shows that in outsourcing, there are at least two positive, or reinforcing, feedback loops which explains why outsourcing projects have a tendency to get out of control: for instance, if trust erodes, parties become less open in their communication, which affects delivery, which in turn further erodes trust, and so on. Of course, the feedback loop can also work in the opposite direction: if trust increases (for instance, by decreasing opportunism), eventually delivery quality will increase, which in turn increases trust further. The point is that the feedback loops make the dynamic aspects of the project explicit. This is the primary benefit of applying the Systems Thinking perspective. 36 IADIS International Conference e-Commerce 2009 The model depicted in Figure 1 follows the so-called qualitative Systems Dynamics paradigm (Sterman, 2000). Like all Systems Thinking paradigms, Systems Dynamics models a system as the interplay between its parts (holism). Systems Dynamics (pioneered by Forrester (1989) and rooted in general mathematical systems theory) focuses on modelling feedback loops consisting of cause-and-effect relations between system components. In qualitative Systems Dynamics, this model is then validated via expert review. In quantitative Systems Dynamics, computer simulations are used to study system dynamics and validate the model. Systems Dynamics primarily deals with quantitative models, but as early as 1983 qualitative approaches have been proposed (Wolstenholme and Coyle, 1983; Wolstenholme, 1983). Figure 1. Operational Success of Outsourcing: Causal Model (van Eck and Ponisio, 2008). A ‘+’ Indicates that an Increase in One Variable Causes an Increase in the Other. A ‘-’ Indicates that an Increase in One Variable Causes a Decrease in the Other. Two ‘Reinforcing’ (Letter ‘R’) Feedback Loops are Indicated Both quantitative as well as qualitative Systems Dynamics are what Pollack (2007) calls ‘hard’ approaches to systems thinking: the systems model is seen as an objective, true representation of the real world, and stakeholders agree on a “clear and single dimensional (single objective)” problem definition (Maani and Cavana, 2000). In ‘soft’ approaches (of which Peter Checkland’s Soft Systems Methodology (Checkland, 1981) is probably the best-known example), the dynamic model is not seen as an objective representation of the real world, but as “a way of generating debate and insight about the real world” (Maani and Cavana, 2000). A ‘soft’ approach has been used by Johnstone et al. (2006) to develop a holistic framework of conflict and conflict resolution in IT projects. 3. PLANNING CO-OPERATION IN INTER-ORGANISATIONAL NETWORKS In this paper we use Systems Thinking to understand inter-organisational networks. In particular, to plan closer co-operation among members of an established inter-organisational network: one formed by customs organisations of the EU. Specifically, network participants need to connect their supporting information systems to improve information flow and to efficiently automatise security checks and paperwork through their borders. Part of the problem is to find ways to help experts (a) find other members that might be interested to co-operate and (b) learn on what topic it is better to collaborate. 37 ISBN: 978-972-8924-89-8 © 2009 IADIS 3.1 Defining Inter-Organisational Networks We define an inter-organisational network as a network of organisations which jointly support value-creating processes (Ponisio, Sikkel, Riemens and van Eck, 2007, 2008). Information flows between the participating organisations. The goal of the network is some result of that flow of information. Each organisation in the network uses a set of information systems (i.e., computer hardware, application software, datasets and possibly manual procedures) to process the information flows mentioned. This set of information systems (the IT portfolio of the organisation) includes both organisation-specific information systems as well as general IT infrastructure components (e.g., email servers). 3.2 eCustoms: an Example of an Inter-organisational Network eCustoms is a representative example of an inter-organisational network as defined in this paper. eCustoms is a network of customs organisations of the 27 member states of the European Union (EU). The network was created in reply to the EU’s aim to facilitate trade and improve ensuring safety of the external borders of the union. Customs organisations of all 27 member states of the European Union have to act as if they form one virtual customs. The network has been in place for many years, but new demands make it necessary for members to co-operate in order to fulfil these demands. 3.3 Existing Approaches to Plan eCustoms The Dutch Tax and Customs Administration conducted between January 2005 and October 2005 a study that compared current organisational context, business processes, systems and future ambitions of ten member states (Netherlands Tax and Customs Administration, 2005). This sample of ten member states is a good one for this paper because it is heterogeneous, representing both large and small countries. This study – called “Benchmarking Customs IT Architecture” – showed the customs networking profiles of each country, providing high-quality data suited to analyse opportunities for co-operation. In fact, this study solved the problem of lack of information required to find opportunities for co-operation. We have read the benchmarking report and used its data as one of the sources of our data collection. The benchmarking study performed by the Dutch Tax and Customs Administration provided a wealth of quantitative data about potential co-operation partners, the analysis of ‘with whom to co-operate’ was performed on an ad hoc basis. To facilitate understanding, in an earlier paper we developed an approach that combines two types of graphical snapshots of the member customs’ relevant properties (Ponisio, Sikkel, Riemens and van Eck, 2008). The result was a visualisation model based on quantitative data that increased understanding of opportunities and challenges in IT integration. A further attempt to optimise planning in eCustoms consisted of using critical problem solving (Ponisio, van Eck and Riemens, 2008). The goal was to present a systematic approach to plan co-operation between customs organisations and the approach was called e-Planning. It consisted in an action plan for customs IT decision makers to decide how to order the steps in the process. e-Planning follows the engineering cycle (Wieringa, 2007) to discover questions that are relevant in a particular scenario. e-planning provided an approach to systematically detect significant co-operation issues, but did not uncover the mechanisms that help decision makers explain the rationale behind their decisions. 3.4 Our Approach: Using Systems Thinking to Extend the Benchmarking Study The previous approaches to improve planning in eCustoms offered solutions to the problem in terms of measuring, visualising and planning, but did not uncover the dynamic mechanisms that explain why some partners are preferred over others. To mitigate this, in this paper we use Systems Thinking to emphasise the dynamics existing in developing inter-organisational networks. This means that we view the eCustoms network as a system, consisting of a number of member states of the EU, their internal organisation, decision making procedures, and supporting information systems. These are the components that interact to create the overall behaviour of the system. In the Systems Thinking approach, we identify properties of these compo- 38 IADIS International Conference e-Commerce 2009 nents as well as relations between these properties. To the best of our understanding our approach is novel: to use Systems Thinking to planning co-operation in inter-organisational networks (specifically in eCustoms) has never been tried before. Specifically, in this paper we use an interpretive case study approach (Klein and Myers, 1999) guided by Systems Thinking. We interpreted the data provided by the benchmarking case study, other documents and interviews with customs experts in terms of the dynamic model presented in Figure 1. We studied the outcome of the benchmarking case study and the other studies looking for similar feedback loops as depicted in the dynamic model, addressing the following two questions: 1. Which data present in the eCustoms case study are causally related to co-operation success potential? This data represents the properties of the underlying system that are of interest. The value of these properties are represented by variables that potentially are part of any feedback loops. 2. What exactly are the causal relations between the variables that we thus uncover? The next section gives an account of how we applied the Systems Thinking view in eCustoms to answer these questions. 4. A CAUSAL MODEL OF PLANNING CO-ORDINATION IN ECUSTOMS In the case study, we applied Systems Thinking to the data collected (e.g., the data provided by “Benchmarking Customs IT Architecture” and the interviews with experts). We reasoned about co-operation planning in a systematic way: we re-interpreted the general causal model (Figure 1) for outsourcing in the context of the eCustoms case study. From this analysis we selected variables to include in the systems thinking model. A variable is included in the model if (i) the variable is involved in an important causal relation and (ii) the variable can be controlled by project managers in real-world situations. The resulting variables and relations together form the eCustoms dynamic model. This model may contain one or more feedback loops, but these feedback loops need not be the same ones as depicted in Figure 1. To the contrary, as the model of Figure 1 is for a situation (software development in outsourcing) that is not the same as in eCustoms, most likely the dynamic model for eCustoms will not be exactly the same. However, the kind of mechanism depicted by the model for outsourcing served as a blueprint for the kind of mechanism that we wanted to uncover in the eCustoms case. The resulting causal model is depicted in Figure 2. The model represents the properties of a potential cooperation between two countries in the eCustoms network. These properties are used to analyse the benefits of co-operation. The properties are case dependent; they are the variables that we derived from documents describing a specific co-operation planning case. In line with Systems Thinking, we view this co-operation as a system. Components of this system are the two participating countries, their goals and needs, their IT systems and data exchange infrastructure, etc. These components have properties that are represented in the model. In the interest of readability, our diagrams have clusters and global arrows. A cluster groups related variables. Our shorthand for having an arrow to or from all the variables in a cluster is to have one arrow to the cluster of variables. The next subsections systematically describe the variables in the model as well as the relations between them. We conclude by explaining concrete examples in terms of the feedback loops in the model. 39 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 2. Systems Thinking Model to Reason about Opportunities and Potential Problems in Planning Co-Operation within the Ecustoms Network 4.1 Cluster 1: Political Support Cluster 1, called ‘Political Support’, consists of three variables that are roughly equivalent to two variables in the general outsourcing model (Figure 1) ‘Trust of O in I’ (top) and ‘Trust of I in O’ (centre). • Matching goals: extent to which the goals in the area of the co-operation topic of the two countries match. • Matching needs: extent to which the needs in the area of the co-operation topic of the two countries are matching (e.g., similar, or complementary) • Image to the world: the extent to which the two countries are able to expose a modern image to the world that shows that they are oriented toward supporting co-operation. These variables are dimensions of the extent to which and decision makers at the political level of the two potential trading partners provide support for the co-operation in terms of allocating resources. They can be interpreted as the ‘willingness’ to co-operate. Thus, political support is broader than just trust. 4.2 Cluster 2: Information Flow Cluster 2, called ‘Information Flow’, consists of two variables that play a similar role in eCustoms as ‘delivery quality perceived by O’ (far right hand side of Figure 1) and ‘Effective coordination’ (just left of ‘delivery quality perceived by O’) in the general outsourcing model. In the eCustoms case study, we focus on quality of information exchange, not on the more general concept of an outsourcer’s perception of project outcome (quality delivered by the insourcer). Therefore, the variables in this cluster in the eCustoms case are more specific: • Member dependence: extent to which the two customs organisations depend on one another (e.g., responsibilities of one are impossible to fulfil without information from the other). • Peer2peer link quality: quality of the information exchange between the two partners. 40 IADIS International Conference e-Commerce 2009 An increase in member dependence and/or link quality causes an increase in political support: increased dependence and link quality represent an investment (sunk cost) that partners want to protect and exploit, which is reflected in matching needs and goals. 4.3 Cluster 3: Operational Potential Cluster 3, called ‘Operational Potential’, contains the remaining six variables of the model that together describe the extent to which necessary assets and procedures are in place at both partners to support close cooperation. • Logistics maturity: Countries differ in how advanced logistics (e.g., transport of goods and import and export procedures) are. This variable represents the difference of logistic maturity of two potential partners. • Architectural knowledge: The difference of architectural knowledge that each country has of its own systems. • Stakeholder commitment: difference in degree of commitment of the stakeholders of each of the two countries to make the link. The stakeholders are the ones mentioned in Ponisio, van Eck and Riemens (2008), i.e., architects, etc. • Stakeholder understanding: difference of the extent of stakeholder understanding of the issues of the system in the other country. • Matching architectures: difference in the extent to which the relevant parts of the IT architecture of each of the two countries is standardised. The potential to match two architectures of the supporting systems of two peer countries (this is at operational level). • Data transfer safety: difference in the extent to which each of the two trading partners ensures safe data transfer and processing. All six variables in this cluster are what we call ‘delta variables’: they represent the difference in quantities of the two potential trading partners, not the quantities for either of them. For all six variables, it holds that an increase in any variable of political support causes a decrease in these differences: the extra resources allocated to the co-operation are used to further standardise IT architectures, implement data security mechanisms, make each potential partner more committed, etc. Moreover, a decrease in this difference causes more and better information flow. 4.4 Examples of Feedback Loops 4.4.1 Dynamics of Political Support As can be seen in Figure 2, the model explains the dynamics of co-operation planning in terms of one balancing feedback loop that comprises all variables in the model. In summary, a decrease in the difference of two countries in terms of operational potential (as represented by the six variables in the cluster with that name) causes better information flow between the partners, which causes an increase in political support. This increase in political support, in turn, decreases the differences between the two countries (additional operational potential), which closes the loop. This process can continue until the delta variables of cluster Operational Potential balance close to zero. Or, if there is an unsolvable mismatch, customs can choose not to cooperate. Making this kind of reasoning early in the planning process is crucial to correct investment decisions because once the co-operation process started and the contracts are signed, it is difficult to leave and to enter the co-operating group. 4.4.2 Example of Late Join-or-quit Feedback Loop As an example of how this feedback loop operates in practice, The Netherlands customs actually did a cooperation with 4 member states on the support and maintenance of NCTS, a software program used to support Transit movements of goods within Europe. Transit movement information systems of individual member states have to comply with the specification of NCTS, but member states are free to choose an implementation. Initially, the Commission provided and supported an implementation of NCTS. About 7 member states (including The Netherlands) used this implementation. In 2007, the Commission announced that they would stop the support of this implementation and member states had to take over. 41 ISBN: 978-972-8924-89-8 © 2009 IADIS Having, thus, a matching need, namely, to support their NCTS system at reduced cost, five member states (including The Netherlands) started a joint project to select a common supplier for the support and maintenance. One member state decided not to continue with the others. The remaining participants decided to make one Request for Proposal to select one common supplier. This was possible because political support and operational potential of this group of four member states was high: they wanted to save cost and there was little difference in architecture, knowledge of each other, etc. After the contract was granted, the four member states started a joint Change Board to discuss changes and future needs with the supplier (information flow). At the end, they replaced the European Commission as a supplier with a commercial vendor. In the contract, they now have more influence on changes than before. The four co-operating member states now have an interest in continuing their co-operation (matching goals), which closes the feedback loop. 4.4.3 Lessons Learnt The dynamics can be explained in terms of the topic of co-operation, the properties of member states that are interested and that actually participate, and the way participants work together over time. The feedback loop in this example suggest the importance of matching goals and needs in the area of the topic of co-operation: if there is not enough benefit in the co-operation (because of the member states properties are changing so there is no match anymore in goals, needs, etc.) a member state can decide not to participate. The example tells us that it is easy to step out in the beginning. Stepping out after the contracts have been signed (exit strategy) has contractual implications, as the contract had exit clauses. In our example, one member could exit easily because the contract included a clause to help in that event. One possible explanation is that the member could have experienced a loss of decision power, because the decisions on a member’s system are constrained by the needs of the group. In our example, participating customs had to agree with others. Similar to leaving a co-operation, it is easier to join in the beginning than after the cooperation has been established, as e.g., important decisions have already been made by other members. In the case of the NCTS example, no member state joined the initial group of four after the contract had been granted. 4.5 Results The results suggest the usefulness of Systems Thinking to plan co-operation in inter-organisational networks. They suggest that Systems Thinking could help decision makers to improve success in finding good partners for closer co-operation, making the search more systematic and serving as a reasoning tool to analyse the underlying mechanisms. The results were discussed with customs experts. The experts found that our approach has potential to help them explain decisions related to choosing the best partner for co-operation. Moreover, according to the experts, applying Systems Thinking (i.e., our approach) can potentially optimise the process of finding good partners for closer co-operation because it systematises the search; which is beneficial compared to the current ad-hoc analysis of with whom to co operate). Furthermore, our approach facilitates reasoning about potential consequences of choosing a given partner early in the process; which happens to be crucial in eCustoms. Regarding internal validity, according to the experts, the method led to the right conclusions (though they would have thought of other variables); which means that the internal validity criterion is met. Our example shows how System Thinking helps stakeholders to reason about the situation and explain the rationale behind decisions. In particular, our model (Figure 2) revealed a feedback loop that (we found later) matched an actual example that occurred in eCustoms. Moreover, applying Systems Thinking to the problem of ‘what should be done first’ in planning cooperation in an inter-organisational network could be generalised to other cases. In fact, the experts could relate to our findings and expressed that our approach helps them in their need to explore new theories that can be used as basis for reasoning, and that foster systematic solutions. Thus, external validity is met. With respect to existing theory, our results are in-line with existing work in the area of co-operation in inter-organisational networks (Finnegan et al., 2001). Specifically, our results concur with the insights of Finnegan et al.: (a) “Inter-organisational systems are based more on the strategies of individual organisations rather than on a network strategy”, and (b) “Planning is a continuous decision activity shared by business and IT”. Moreover, our method remains consistent with previous work in the area of power relationships in net- 42 IADIS International Conference e-Commerce 2009 works (Emerson, 1962). In particular, our approach reveals power dependence relations. For instance, the concrete real-life example we present shows that participants ended having more influence on the changes (to their common system) than before. In addition to being in-line with existing research, our approach goes a step further than just explanation by providing a holistic solution to a concrete and complex problem. 4.5.1 Future Work Possible extensions to the model include adding more variables. The literature of eCustoms provides numerous suggestions; see (Netherlands Tax and Customs Administration, 2005) for qualitative and quantitative data supporting reasoning about potential clusters in eCustoms. In this line of reasoning, candidate properties are organisational autonomy, properties of the current IT architecture portfolio and transaction volumes. Another type of extension is related to our definition of success: our model focuses on operational success in the event of making two customs’ systems co-operate (short term). However, collaboration success is related to the way network participants work together not only in the short-term, but also in the long run. Whether it is possible to extend the model with consideration to explicit evolution is a topic for future research. In a different line of reasoning, the validation of our approach suggests a new path for future research. Specifically, systems thinking does not provide a systematic way to find the variables of a dynamic model. Textbooks on systems thinking only advise to organise a brainstorming session with subject matter experts. A possible direction for future work is to investigate whether automatic techniques empowered by visualisation prove to be efficacious to discover the variables. 5. CONCLUSION In this paper, we applied a Systems Thinking approach to study the problem of selecting partners for closer co-operation in e-commerce. We uncovered a feedback loop that explains how an increase in political support for a potential partner (thanks to e.g., matching needs and goals) causes (via the creation of more operational potential) an increase in information flow, which in itself increases political support. The feedback loop can also operate in a downward way: a decrease in operational potential causes (via a decrease in information flow) a decrease in political support, which in turn further decreases operational potential. The results suggest the usefulness of Systems Thinking to plan co-operation in inter-organisational networks. The mechanism uncovered was confirmed by experts and by a concrete example. The practical implication of uncovering this mechanism is that it enables stakeholders to improve the partner selection process, supporting understanding of the forces and tensions that govern partner selection. The mechanism also explains why it is important to select potential partners for closer co-operation as early as possible, the importance of willingness to participate, and the gain or loss of decision power that joining a network implies. ACKNOWLEDGMENTS We gratefully acknowledge the financial support of the Netherlands Organisation for Scientific Research (Dutch Jacquard program) for the project 638.004.609 (QuadRead). REFERENCES Checkland, P. (1981), Systems Thinking, Systems Practice, John Wiley & Sons Ltd. Emerson, R. M. (1962), “Power-dependence relations”, American Sociological Review, Vol. 27, pp. 31–41. Finnegan, P., Galliers, R. D. and Powell, P. (2001), “Operationalising guidelines for inter-organisational systems planing: Exploring a learning model [research in progress]”, in 9th European Conference on Information Systems, Bled, Slovenia, June 27-29. 43 ISBN: 978-972-8924-89-8 © 2009 IADIS Forrester, J. W. (1989), The beginning of system dynamics, Working paper D-4165-1, http://sysdyn.clexchange.org/sdep/papers/D-4165-1.pdf Sloan School of Management, Massachusetts Institute of Technology. Gerst, M. and Bunduchi, R. (2005), “Shaping IT standardization in the automotive industry – the role of power in driving portal standardization”, Electronic Markets, Vol. 15 No. 4, pp. 335–343. Gerst, M. and Bunduchi, R. (2007), “Intraorganisational power and the adoption of interorganisational IT innovations: the inside story of Covisint”, International Journal of Technology Intelligence and Planning, Vol. 3 No. 1, pp. 57–74. Johnstone, D., Huff, S. and Hope, B. (2006), “IT projects: Conflict, governance, and systems thinking”, in Proc. 39th Annual Hawaii International Conference on System Sciences HICSS ’06, Vol. 8, pp. 197b–197b. DOI: 10.1109/HICSS.2006.235 Klein, H. K. and Myers, M. D. (1999), “A set of principles for conducting and evaluating interpretive field studies in information systems”, MIS Quarterly, Vol. 23 No. 1, pp. 67–93. Maani, K. E. and Cavana, R. Y. (2000), Systems Thinking and Modelling: Understanding Change and Complexity, Prentice Hall, Auckland, New Zealand. http://books.google.com/books?id=IF-YAAAACAAJ Midgley, G. editor (2003), Systems Thinking, SAGE,. http://books.google.com/books?id=nCp7HQAACAAJ Netherlands Tax and Customs Administration (2005), “Benchmarking customs IT architecture. exploit commonality, manage variability”. Pollack, J. (2007), “The changing paradigms of project management”, International Journal of Project Management, Vol. 25 No. 3, pp. 266–274. DOI: 10.1016/j.ijproman.2006.08.002 Ponisio, M. L., Sikkel, K., Riemens, L. and van Eck, P. (2007), “Understanding Cooperation in Inter-Organisational Systems”. In: Proceedings of the International Association for Development of the Information Society (IADIS) International Conference e-Commerce 2007, 7-9 December 2007, Algarve, Portugal. pp. 163-174. IADIS Press. Ponisio, M. L., Sikkel, K., Riemens, L. and van Eck, P. (2008), “Combining visualisation techniques to understand cooperation in inter-organisational systems”, Journal of Systems and Information Technology, Vol. 10 No. 2, pp. 159– 179. DOI: 10.1108/13287260810897774 Ponisio, M. L., van Eck, P. and Riemens, L. (2008), “Using critical problem solving to plan inter-organisational cooperation in e-customs”, in Proceedings of the International Association for Development of the Information Society (IADIS) International Conference e-Commerce 2008, Amsterdam, The Netherlands, IADIS Press, pp. 28–39. Sterman, J. D. (2000), Business dynamics: systems thinking and modeling for a complex world. Irwin/McGraw-Hill. van Eck, P. and Ponisio, M. L. (2008), “IT project management from a systems thinking perspective: A position paper”, in Proceedings of the 3rd International Research Workshop on IT Project Management (IRWITPM 2008) part of the International Conference on Information Systems (ICIS 08). Wieringa, R. (2007), Research and design methodology for software and information engineers. www.cs.utwente.nl/~roelw Internal report. Williamson, O. E. (1993), “Opportunism and its critics”, Managerial and Decision Economics, Vol. 14 No. 2, pp. 97– 107. Wolstenholme, E. F. (1983), “Modelling national development programmes – an exercise in system description and qualitative analysis using system dynamics”, The Journal of the Operational Research Society, Vol. 34 No. 12, pp. 1133–1148. http://www.jstor.org/stable/2581837 Wolstenholme, E. F. and Coyle, R. G. (1983), “The development of system dynamics as a methodology for system description and qualitative analysis”, The Journal of the Operational Research Society, Vol. 34 No. 7, pp. 569–581. http://www.jstor.org/stable/2581770 44 IADIS International Conference e-Commerce 2009 FINDING TREND LEADERS FROM MARKETING TRANSACTION DATA Masakazu Takahashi, Kazuhiko Tsuda Graduate School of Business Sciences University of Tsukuba 3-29-1, Otsuka, Bunkyo, Tokyo 112-0012, Japan Takao Terano Dept. Computational Intelligence and Systems Science Tokyo Institute of Technology 4259-J2-52 Nagatsuda, Midori-ku, Yokohama 226-8502, Japan ABSTRACT This paper presents a new recommendation method in marketing retails from the transaction data, which may contain frequent customers’ consumption activities. Contrary to conventional popular engines, which generates recommendation the results from the information by major customers such as among such as e-Bay does, we have formulated indices of trend leaders with the criteria for the purchased date and the number of sales in order to get the good-to-sale items, which will increase the number of the sales in the future. We have confirmed that it is possible to make appropriate recommendations to the other customers from the transitions of preferences of trend leaders in the following senses: 1) the effect of the recommendation with the trend leaders’ preferences, 2) the extraction of good-to-sale marketing items from the frequent customers’ transactions, and 3) the early purchase effects on the short best-before and the long bestbefore date items. KEYWORDS Recommendation Systems, Collaborative Filtering System, Customer Preferences, Data Analyses 1. INTRODUCTION The inefficiency of retail store operations is caused by several reasons, but mainly the misunderstandings of the timing of customers’ needs and the preferences. If the marketing items, which a certain customer prefers to, were out of stock, it would be nonsense to generate recommendations for them. It is important to know customers’ latent needs and preferences. The issue would affect the whole retail industry including manufacturers of sales items. We need to properly and automatically obtain the both the preferences and the timings for the customers. Concerning to the understanding for the customers’ preferences, a collaborative filtering system is commonly used among the B2B or the B2C industries [Netperceptions2000]. Whereas, their systems often only utilize the following information: purchasing dates, name of the items, customer IDs , number of the sold items, and the price of them. These attributes are stored in the retail POS (Point-OfSales) systems with the reward point cards, which will identify the customer and the transactional machines, and the purchased items. Although POS systems are widely adapted even to the small retail stores in Japan, there have been few attempts to integrate these data to understanding consumer behaviors. Based on the background, in this paper, we present a new a new recommendation method in marketing retails from the transaction data, which may contain frequent customers’ consumption activities. The rest of the paper is organized as follows: Section 2 discusses the related work and weak points of the state-of-the-art recommendation systems; Section 3 describes how we extract important information from retail transaction data; Section 4 evaluates the proposed method, and finally Section 5 gives some concluding remarks and our future work. 45 ISBN: 978-972-8924-89-8 © 2009 IADIS 2. RELATED WORK Although recommendation systems are in common among the B2B or B2C businesses [Hijikata2007] [Linden2003] [Orma2006], the conventional recommendation engines only provide the proper results for commonly sold items and typical customers. They will show the weakness for both newly developed items and new comers, when we would apply the methods to the retail industry. The customers of the retail industry should be categorized such divergent points as their preferences, incomes, and the number of family, and so on. We must care the recommendation information taking account of both the properties and their diversity. 2.1 Understanding for the Customer Preferences Collaborative filtering is one of the tools to understand the customer preferences and has the characteristics with easiness to application and the affordability for the huge data. On the other hand, the system with collaborative filtering is often influenced by the majority answers of the items and the customers [Adomavicius2005] [Bruke2002] [Herlocker2004]. Therefore, collaborative filtering requires a large amount of information related to items and consumers: a large amount of the information about the items and the users must be categorized into appropriate segments. The feature comes from the method to group the customer preference with the similarity [Netperceptions2000] [Schafer2001]. The steps for generating the recommendation information are summarized as follows; (1) correct many purchase history data from customers, (2) make appropriate customer groups, which purchased the same or similar items, when a certain customer purchased the item, and (3) generate the recommendation information based on the item groups among the same group of customers. In collaborative filtering, they do not analyze the contents of items at all. Therefore, it restricts the recommendation objects only purchased ones, and it generates customer's community only from the purchase information without the analysis of contents of the item. Such shortcomings of the collaborative filtering systems are described in [Denning2006] [Herlocker2004]. Denning, et al. noted that the collaborative filtering system is required both the large number of customers and the amount of contents. Therefore, the same recommendation results will be generated if the items to be recommended are little. Moreover, the existing items only are recommended. This is a critical defect when we must sell new release items. As for the forecasting the number of sales of new items, Nakamura has evaluated and classified the characteristics of the new items with the market reflection data such as from the POS data with customer IDs [Nakamura2001]. The conventional evaluation methods such as the trial repeat model only indicate the characteristics for the items which will be purchased repeatedly, but are insufficient for the customer classification or the recommendation from the ID-POS data classification [Abe2005]. 3. METHOD TO PROCESS TRANSACTION DATA One of the conventional methods for the timing understanding on the customer needs is the time-series data analyses with the quantitative data such as the POS data. The method is good for the inventory controlling such as the stable demands items from the customers needs, because of the sufficient time-series data and able to mark high score ratio for the demand forecasting. On the other hand, it is hard to forecast the demands on such as the new items or the items that demand rapidly stood up, because of the insufficient of the forecasting based data. This section proposes the algorithm to efficiently detect the timing of the item needs from the ID-POS data. Especially, to understand the timing of the item needs, we make use of the concept of the trend leaders to forecast the demand of new items. 3.1 Acquiring ID-POS Data The ID-POS system is commonly used in the Japanese retail industry, even they are so small. We have corrected ID-POS data from a local grocery store located at Shimane prefecture in Japan between from May to December in 2007 in order to find items to be sold in the future. Table.1 summarizes transaction 46 IADIS International Conference e-Commerce 2009 information acquired from the gathered data. They have about 10,000 customers visited with about 70,000 transactions during the observation period. There are about 20,000 items registered in the POS system with the variety of the conventional items, the seasonal items, and new released items. Table 1. Summary of the Number of Transaction Information Transactions Customers Purchased Items May Jun Jul Aug Sep Oct Nov Dec 69,106 69,681 68,853 69,275 69,294 71,403 67,182 71,192 9,766 9,970 9,791 9,832 10,021 9,980 9,872 10,279 689,592 689,686 658,716 674,667 680,102 690,444 651,723 705,600 Table.2 describes the number of the new registered items during the period. There are the 1,922 items registered. Among them, the 1,738 items were sold at least one piece after the release and among the 1008 items have been increased the sales number from the release to 28 days later. Those who purchased the 1,008 items amount the 9,141 customers. Table 2. Transaction Attributes Items in POS New Registered Purchased Items Increased Items ActiveCustomers 18,091 1,922 1,738 1,008 9,141 Table. 3 shows the relation between the number of items of the new registered-, the purchased-, and the increased-items. The table also shows the percentage of the purchased items and the sales increased items. From the figure, about 50% of the new registered items could not increase the sales number. As for September, the 425 items has registered because of the seasonal change from summer to fall. Table 3. Number of Items for the Registered, the Purchased, and the Sales Increased May Jun Jul Aug Sep Oct Nov Dec Number of Registered 192 250 226 215 425 284 191 139 Purchased Items 161 230 198 195 376 256 183 139 Include Increased 91 123 99 119 284 130 100 62 3.2 Extraction Algorithm To extract the items to be sold, we have found the indices of trend leaders who bought the items at the initial stage. They are expected to maximize the sales amount in the future. We would like to observe the activities of those customers. To measure the indices of the trend leaders for each item, we determine the following criteria: 1) How first they have purchased the items from the release points, and 2) how many purchased items would increase the sales number after the period. The procedure to find the items to be sold from the ID-POS data is summarized in the following steps: 1. Item based calculation Step1 Count the date for the release Step2 Count amount of sales for 28days from the release Step3 Count amount of sales for 35days from the release Step4 Subtract the result of Step3 from the result of Step2 Step5 Extract the items of Step 4 > 0 Extraction with high support items from the customers from the amount of the sales 2. Customer based calculation Step6 Count the number of the sales Step7 Correct the purchased date for each item Step8 Subtract the date for start selling from the purchased date for each item Calculate the index of the early purchase Step9 Summation for the reciprocals of the early purchase date For the calculation of the early purchase function Step10 Count the items of the step5 among the step 6 47 ISBN: 978-972-8924-89-8 © 2009 IADIS Step11 Calculate Step9 / step6 Table.4 indicates the demographic information of the indices of the trend leaders (ITL) based on the above procedure. From the table, the maximum of the ITL was 6.91, and minimum ITL was 0.00. 8,707 among 9,141 customers show 1.00 or less for the index. The customers exceeded 6.00 or more were only 3 persons. Table 4. Demographic Information of the Indices of Trend Leaders Index ~1 ~2 ~3 ~4 ~5 ~6 ~7 total Customer 8,707 360 45 20 4 2 3 9,141 Percentage 95.25% 3.94% 0.49% 0.22% 0.04% 0.02% 0.03% 100% Fig.1 shows the relation between the number of the purchased items and the index of the trend leaders. Most of the customer purchased less than 50 items and scored the low index of the trend leader. Otherwise, not always scored high index of the trend leader but purchased lot. For example, those who scored more than 4.00, the number of the purchased items varied wide. 7 y = 1E‐07x3 ‐ 8E‐06x2 + 0.0264x + 0.0189 R² = 0.7425 6 Index of the Trend Leader 5 4 3 2 1 0 0 20 40 60 80 100 120 140 160 180 200 Number of the Purchased items Figure 1. Number of the Purchased Items and the ITL The figure suggests that, if we set the temporary threshold of the trend leader scored more than two, then we can understand the following phenomena: 1)Except the exceeded score for the trend leader, those who scored high index usually locates the near the store. 2) Almost all the locations have the trend leader. From the relation between the demography of the ITL and the other attribute, depending on the threshold for the trend leader, we can use the criteria as the key index for the recommendation based on the preference and the timing for the needs. 4. RECOMMENDATION ENGINE EVALUATIONS The index of the trend leaders is one of the reference values, therefore, it is necessary to set the threshold of the index properly according to characteristics of the region, items, customer attributes, and sales policy. Then, index of more than 2 is assumed to be a threshold, both to extract the trend leader and to recommend the items. As for the recommendation engine, in this paper, we took the Taste; open source software based on JAVA [Taste]. This engine is originally used the correlation coefficient for the similarity calculation method, to hold many indexes of the similarity, the similarity method was changed to the cosign distance in this thesis. 48 IADIS International Conference e-Commerce 2009 It is difficult to detect the ability with the trend leader of the item selection that foreseen the fashion in advance from the numerical data such as the ID-POS transactional data. Moreover, the candidates for the trend leader include the customer only with short span curiosity, so that an efficient selection method of detecting the trend leader is required. 4.1 Evaluation for the System with the Short Best-before Date Item This section, we evaluate the recommendation performance with the item of new released milk that obtains the short best-before date at least one week or so. This item is labeled number for 4902081205724 based on the JAN (Japanese Article Number) regulation. This unique code is issued by each company and searched by the GEPIR (the Global Electronic Party Information Register) site [GEPIR]. This unique number is composed of the following attributes in case of 13 digits; 1) country code for 2 digits, 2) company code 7 digits, 3) item code for 3 digits, and 4) chick digit for 1 digit. For this unique identification number enable us to identify the transaction of the item. For example, applying to this classification to the item of the simulation, 49 for the country code of Japan, 02008120 for the company code, 572 for the item code, and 4 is for the check digit respectively. With the 74 customers scored more than 2 of the index of the trend leader, we obtain the following results with the 10-fold cross-validation. Fig.2 indicates the relation between the increase ratio for the item and the index of the trend leader. This indicates the performances of the proposed index. From the graph, we obtained that the more scored of the index of the trend leader, the more scored high increase ratio for the gross sales number except the around scored 5. This result of the exception comes from the preferences of the each customer. This result obtains the slightly different performances with the customer preferences and shows the diversity for the preference of the customers. Increase ratio for the Gross Sales Number(%) 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 Index of the Trend Leader Jun Jul Aug Sep Oct Nov Dec Figure 2. Increase Ratio for the Sales Number (Short Best-before Date Item) Fig.3 indicates the relation between the early purchase effect and the index of the trend leader. This also indicates the performances of the proposed index with the early purchase effect. From the graph, we obtained the following issue; the more scored of the index of the trend leader, the more scored earlier. This also obtains the slightly different performances with the customer preferences. This means the diversities for the preferences of the customers. 49 ISBN: 978-972-8924-89-8 © 2009 IADIS 7.0 6.0 Early Purchase Date 5.0 4.0 3.0 2.0 1.0 0.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 Index of the Trend Leader Jun Jul Aug Sep Oct Nov Dec Figure 3. Early Purchase Effect (Short Best-before Date Item) Table.5 and Table.6 indicates the recommendation results of the item with the proposed index. As for the increased ratio for the item sales number scored between 0.13% and 1.60% rise wit the average of 0.44% and the early purchase effect marked between 0.62 day and 6.61 days with the average of 1.71 days earlier than the default. The recommendation information from the trend leaders with high score made increase the sales number and early purchased days. We found out the items that will become the high potential for the increasing sales number in the future both the proposed index and the short best-before date item. Table 5. Recommendation Results for the Increase Ratio (%) (Short Best-before Date Item) Max Min Average Standard Deviation 1.60 0.13 0.44 0.23 Table 6. Recommendation Results for the Early Purchase Effect (Days) (Short Best-before Date Item) Max Min Average Standard Deviation 6.61 0.62 1.71 0.84 4.2 Evaluation for the System with the Long Best-before Date Item Next, for the performance evaluation of the system, we took the item of new released pops that obtains the long best-before date at least over month labeled number for 4902179011725 with the same conditions. Fig.4 indicates the relation between the increase ratio for the item and the index of the trend leader with the long best-before date item. From the figure, we found out the item with the long best-before date item could not affect the sales increase. Fig.5 also indicates the relation between the early purchase effect and the index of the trend leader with the long best-before date item. From the both graph, we obtained from the simulation that the more scored of the index of the trend leader, the more scored early purchase effect and the low effect for the increase ratio for the item with the long best-before date item. 50 IADIS International Conference e-Commerce 2009 Increase ratio for the Gross Sales Number(%) 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 Index of the Trend Leader Jun Jul Aug Sep Oct Nov Dec Figure 4. Increase Ratio for the Sales Number (Long Best-before Date Item) 8.0 7.0 Early Purchase Date 6.0 5.0 4.0 3.0 2.0 1.0 0.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 Index of the Trend Leader Jun Jul Aug Sep Oct Nov Dec Figure 5. Early Purchase Effect (Long Best-before Date Item) Table.7 and Table.8 indicates the recommendation results of the item with the proposed index. As for the increased ratio for the item sales number scored between 0.09% and 0.82% rise wit the average of 0.29% and the early purchase effect marked between 0.74 day and 7.44 days with the average of 2.03 days earlier than the default. The recommendation information from the trend leaders with high score made early purchased days effect. We found out the items that will become the high potential for the early purchase effect with both the proposed indexes and the long best-before date item. Table 7. Recommendation Results for the Increase Ratio (%) (Long Best-before Date Item) Max Min Average Standard Deviation 0.82 0.09 0.29 0.14 Table 8. Recommendation Results for the Early Purchase Effect (Days) (Long Best-before Date Item) Max Min Average Standard Deviation 7.44 0.74 2.03 1.01 51 ISBN: 978-972-8924-89-8 © 2009 IADIS As the benefits for the recommendation with the proposed indexes, we have succeeded detect the items that will increase sales number in the future, and extract the customer groups that became the trend leaders’ candidates especially to good for the short best-before date items. 5. CONCLUDING REMARKS In this paper, we have described a new method to analyze the ID-POS transactional data. From the experiments, we have found out a formula to extract the trend leaders among the customers. We have confirmed that 1) we are able to make appropriate recommendations to the other group members based on the transitions of the trend leaders' preferences, 2) from the trend leaders among the customers, we are able to make appropriate recommendations to the other group members based on the transitions of the trend leaders' preferences, and 3) the effect of the recommendation with the trend leaders’ preferences. Using the results, we will make detailed decisions in the following three points: 1) to make appropriate recommendations to the other group members based on the transitions of the trend leaders' preferences; 2) to evaluate the effects of the recommendations with the trend leaders’ preferences; and 3) to improve the retail such management processes as prevention from the out of stock phenomena, sales promotion for early purchase effects, and the increase of the numbers of sales. Furthermore, to make use of the indices, we will evaluate items from both of the long best-before date and the short best-before date, in order to find out the short best-before date item is good for sales expansion. This research is supported in part by a local super market in Japan. We express appreciation to those involved. REFERENCES [Abe2005]Abe, M., and Kondo, F., 2005, Science of Marketing, POS Data Analysis, Asakura Publishing, Tokyo, Japan (Japanese) [Adomavicius2005]Adomavicius, G., and Tuzhilin, A., 2005, Toward the Next Generation of Recommender Systems: A Survey of the State-of-the-Art and Possible Extensions, IEEE Trans. on Knowledge and Data Engineering, Vol.17, No.6, pp.734-749 [Bruke2002]Burke, R., 2002, Hybrid Recommender Systems: Survey and Experiments, User Modeling and UserAdapted Interaction, Vol.12, pp.331-370 [Denning2006]Denning, P.J., and Dunham, R., 2006, The Missing Customer, Communications of the ACM, Vol.50, No.4, pp.19-23 [GEPIR]http://www.gepir.jp [Herlocker2004]Herlocker, J. et al, 2004, Evaluating Collaborative Filtering Recommender Systems, ACM Transactions on Information Systems, Vol.22, No.1, pp.5-53 [Hijikata2007]Hijikata, Y., 2007, Techniques of Preference Extraction for Information Recommendation, Journal of Information Processing Society of Japan, Vol.48, No.9, pp. 957-965 [Linden2003]Linden, G. et al, 2003, Amazon.com Recommendations; Item-to-Item Collaborative Filtering, IEEE INTERNET COMPUTING, Jan-Feb, pp.73-80 [Orma2006]Orma, L.V., 2006, Consumer Support Systems, Communications of the ACM, Vol.50, No.4, pp.49-54 [Nakamura2001]Nakamura, H., 2001, Marketing of the New Products, Chuokeizai, Tokyo, Japan (Japanese) [Netperceptions2000]NetPerceptions, Recommendation Engine White Paper, 2000 http://www.netperceptions.com/literature/content/recommendation.pdf [Schafer2001]Schafer, J.B. et al, 2001, E-Commerce Recommendation Applications, Data Mining and Knowledge Discovery, Vol.5, pp.115–153 [Taste]http://taste.sourceforge.net 52 IADIS International Conference e-Commerce 2009 ADDING VALUE TO ENTERPRISEWIDE SYSTEM INTEGRATION: A NEW THEORETICAL FRAMEWORK FOR ASSESSING TECHNOLOGY ADOPTION OUTCOMES Linda Wilkins RMIT University, Melbourne, Australia Paula M.C. Swatman University of South Australia Duncan Holt Raytheon Australia ABSTRACT The concept of an enterprise wide information system (EWS) strategy is often invoked as a means of repositioning an organisation to do what it does best. Streamlining processes and integrating systems in key organisational areas such as records management is fundamental to such a strategy and has led to growing uptake of Electronic Document and Records Management Systems (EDRMS). During the implementation, a project manager will need an enterprise model or framework that accurately captures the reality of the organisation and its operational tasks and assists monitoring of the project to minimise risk. However in the IS field no single theoretical framework or model has gained widespread acceptance. In this paper we draw on Evolutionary Diffusion of Innovation Theory (EDT) and TAM/TAM2/UTAUT to evaluate their effectiveness as predictors for a successful EDRMS implementation in a local government context. We report on modifications to these two frameworks that enabled the development of a technology implementation and acceptance (TIAM) model. The new model enables improved measurement and monitoring of EWS implementations as they occur. KEYWORDS Diffusion theory; Technology Acceptance Model; case study; SMEs 1. THEORETICAL FRAMEWORKS FOR TECHNOLOGY ADOPTION IS implementation studies are in need of a widely accepted theoretical framework satisfactorily explaining how complex and networked technologies diffuse. Paradigms applied to implementation studies in the IS field frequently draw on two bodies of theory or constructs – Diffusion of Innovation (DoI) and the Technology Acceptance Model (TAM). We begin with an outline of the elements of each theory or construct, consider the benefits of applying these to studies of IS process innovations and conclude with some findings on the limitations of each theory’s explanatory power. 1.1 Diffusion of Innovation/Evolutionary Diffusion Theory Diffusion of innovation theory attempts to explain how an innovation is spread and why it is adopted. Evolutionary Diffusion Theory (EDT) deals with the differentiations, complexities and uncertainties existing in the overall economic structure (Lambooy and Boschma, 2001; Lissoni and Metcalfe, 1994). EDT seeks to explain the factors that often lead to sub-optimal behaviour in the real world – particularly where decisionmaking is complex and the decision-making environment contains multiple interdependencies (Arthur 1994). Key features of EDT include: rejection of optimisation or the feasibility of determining one ‘best’ policy; the role of government as policy maker; the acceptance of human intervention in economic processes; and a 53 ISBN: 978-972-8924-89-8 © 2009 IADIS focus on systems and markets rather than individual firms. The first three of these EDT axioms (described below) can be applied to analyses at both sectoral and organisational levels. Rejection of Optimisation: EDT rejects the idea of optimisation or implementing one ‘best’ policy (Metcalfe, 1994). Successful innovations represent the outcome of multiple and contingent variables and do not always have to be the best ones (Saviotti and Metcalfe, 1991). Models based on Evolutionary Diffusion theory thus stress the gradualism of internal adoption. Government as Policy Maker: The role policy intervention can play and the institutional pressure government can exert to stimulate innovative technology uptake were explored by Lissoni and Metcalfe (1994); and later extended to understanding the role of government as policy maker by Lambooy and Boschma (2001), who emphasise the need for technology solutions to be context-specific and sensitive to local path dependencies. Human Intervention in Economic Processes: Implicit in Evolutionary Diffusion Theory is the assumption that intervention in the process of technology development is possible and that the selection of a theory can influence policy design. Once the soft components of technology innovation are recognised, actors are clearly capable of consciously attempting to change their environment (Nelson 1995). As a theoretical framework, EDT extends the explanatory power of diffusion theory and offers a basis for understanding many of the complex and interrelated issues applying to instances of e-business uptake. Application: Evolutionary Diffusion Theory moves on from a focus on the innovation itself to integrating the actors involved, the firm and the social environment. As complementary theories, they provide considerable explanatory strength for studies of process innovation. As a theoretical framework EDT should be applicable to all stages of a technology implementation. Limitations: Evolutionary Diffusion Theory is focused on systems and markets rather than individual firms. Its orientation towards inter-organisational studies of industry sectors and markets may make it less applicable to studies of single organisation implementations. 1.2 Technology Acceptance Model (TAM) Davis, Bagozzi and Warshaw’s (1989) Technology Acceptance Model (TAM) is one of the most influential theories in IT/IS adoption behaviour, adapting Ajzen and Fishbein’s (1980) Theory of Reasoned Action (TRA) to explain why users accept and use technology, as well as the influence factors involved in these processes. The limited association between a system's acceptance and its usage intensity was the main driver for the development of TAM, which was designed to create a foundation for research into the impact and influence of external variables on internal beliefs, attitudes and intentions towards technology acceptance (Davis et al., 1989). Elements: Perceived usefulness and perceived ease of use determine an individual's intention to use a system, with ‘intention to use’ serving as a mediator of actual system use. Perceived ease of use influences perceptions of usefulness, though Davis (1986) found usefulness was far more strongly linked to usage than ease of use. TAM assumes intention to use an IS is free of other individual, organisational and environmental constraints (Bagozzi et al., 1992). The theory utilised a set of external variables (e.g. individual abilities, tasks, or types of system) by which a user could determine the implications of using an IS (the Subjective Norm), informing his/her beliefs about perceived usefulness and ease of use of the system, creating the user’s attitudes towards this particular IS – which, in turn, would drive his/her behavioural intention towards actual system use (Burton-Jones and Hubona, 2005; Davis et al., 1989). The limitations of the original TAM led Venkatesh and Davis (2000) to develop TAM2, an extension which simplified the originally complex chain of IS user logic by eliminating the Attitude Toward Using variable and clarifying ‘Perceived Usefulness’ and ‘Intention To Use’ via the provision of two new groups of influences (Social Influence Processes and Cognitive Instrumental Processes) clarifying the user’s motivation(s) for using the system. Venkatesh and Davis (2000) identified three separate Social Influence processes – Voluntariness, Subjective Norm and Image. • Voluntariness describes the degree to which potential users understand the adoption decision to be non-compulsory (Agarwal and Prasad, 1997). Usage intention towards an introduced IS is likely to vary depending on whether use of the system is voluntary or mandatory. An individual’s willingness or unwillingness to comply with such mandates affect intention to use the system (Venkatesh and Davis, 2000). 54 IADIS International Conference e-Commerce 2009 • The Subjective Norm represents ‘a person’s perception that most people who are important to him think he should or should not perform the behaviour in question’ {Fishbein and Ajzen, 1975, p302}, and directly influences intention to use the system. • TAM2 uses Moore and Benbasat’s (1991) definition of Image as 'the degree to which use of an innovation is perceived to enhance one’s…status in one’s social system' (p.195) and proposes a direct correlation between Image and Perceived Usefulness. Experience is another important factor in TAM2, although Venkatesh and Davis (2000) did not categorise experience as a social influence process, instead keeping this variable separate – but relating it to the Social Influence group of processes. The Subjective Norm directly influences Intention to Use in the early stages of a mandatory system’s implementation within an organisation – but the influence of the Subjective Norm on Intention to Use will decrease over time and will instead be replaced by Experience of using the system (Venkatesh and Davis, 2000). The group of Cognitive Instrumental Processes in TAM2 includes four variables: Job Relevance, Output Quality, Result Demonstrability and Perceived Ease of Use, with the first three directly affecting Perceived Usefulness: • Job Relevance relies on the user's judgment on the extent to which an IS is applicable to his/her own work; • Output Quality represents a system’s performance in relation to its given tasks and affects Perceived Usefulness (see Davis et al., 1989) • TAM2 utilises Moore and Benbasat’s (1991) definition of Result Demonstrability as the ‘tangibility of the results of using the innovation’ (p. 203); • Perceived Ease of Use is equal to Perceived Usefulness and Intention to Use in terms of its importance as an influence on a user’s decision to utilise the IS. Application: As a general purpose model of IT adoption, the TAM/TAM2 models provide significantly improved insight. These models are particularly applicable where evaluation of the usefulness of an innovative ISD process is required. Limitations: Both TAM and TAM2 assume that intention to act implies limitless freedom. In practice, however, constraints such as limited ability, time, environmental or organisational limits, and unconscious habits will limit this freedom. Constructs based on TAM also focus on perceived personal benefits of the innovation but fail to consider organisational benefits (Hardgrave and Johnson 2003 p.326). It is widely acknowledged that TAM (unlike EDT) does not explain why performance beliefs often disagree with objective reality (Davis et al, 1989, p.335). More recent studies have also found that ‘certain cultures are especially sensitive to user perceptions, and practitioners should not assume that perceptions of ease of use and usefulness are universal’ – a limitation likely to have some impact on IT out-sourcing decisions (McCoy et al 2007:p.89). 1.3 UTAUT (Unified Theory of Acceptance and Use of Technology) The UTAUT model developed by Venkatesh et al. (2003) – which, like TAM/TAM2, is focused on the description and explanation of organisational acceptance of a technology (Carlsson 2006) – is based on a review of 8 prominent models1 in user acceptance research. It provides insight into both the intentions of potential adopters to use an IS, as well as their usage behaviour. The UTAUT model has its four key constructs: performance expectancy, effort expectancy, social influence and facilitating conditions which directly influence behavioural intention and use behaviour. Gender, age, experience and voluntariness of use are key moderators for the impact each construct has on behavioural intention and use behaviour. These comparatively brief introductions to Evolutionary Diffusion Theory and the Technology Acceptance Model body of work which culminated in UTAUT illustrate the very different perspectives these theories provide on technology innovation uptake: EDT is essentially focused on the organisation itself rather than on determining individual attitudes towards IS use, while TAM/TAM2 and UTAUT are largely focused on individual attitudes towards IS adoption rather than organisational influences/impacts on individual attitudes to IS innovation. Clearly, both approaches have a role to play in forecasting the likely success of any 1 Theory of Reasoned Action (TRA), Technology Acceptance (TAM, TAM2), Motivational Model (MM), Theory of Planned Behaviour (TPB), Model of PC Utilization (MPCU), Combined TAM-TPB, Innovation Diffusion Theory (IDT), Social Cognitive Theory (SCT) 55 ISBN: 978-972-8924-89-8 © 2009 IADIS specific IS implementation. Table 1 compares the basic assumptions of these three theories, their strengths and limitations. Table 1. Comparative Assessment of EDT, TAM/TAM2 and UTAUT Theoretical Foundation Evolutionary Diffusion Theory Basic Assumptions Stresses the gradualism of internal adoption Presents adoption of single innovations as part of a greater process of change affecting organisations and their culture TAM/TAM2 UTAUT Recognises that actual usage may not be a direct or immediate consequence of attitudes and intentions Focuses on the way in which determinants of intention and behaviour evolve over time Highlights the role and importance of contextual analysis when developing strategies for organisational technology implementations Strengths Limitations Recognises the possibility of changes in an innovation during the adoption and implementation process More applicable to interorganisational studies and markets than single organisation studies Provide guidance as to how to accelerate the rate of adoption and predict outcomes Replaces multiple attitude measures with two validated technology acceptance measures— ease of use, and usefulness. Does not consider the benefits to the organisation Unified but parsimonious model which, building on eight previous models, potentially offers the benefits of all Refines the body of work which culminated in TAM Ignore the organisational/ personal perspective Does not consider significant user determinants such as ‘intrinsic motivation’ or ‘attitude toward behaviour’ Does not consider the impact of attitudes to using a particular technology innovation In the next section of this paper we review an EDRMS implementation in a local government authority. We then apply EDT, TAM/TAM2 and UTAUT to the case study and assess the relative explanatory strength and/or limitations of each of these approaches. 2. A LOCAL GOVERNMENT EDRMS IMPLEMENTATION An EDRMS implementation with fully scanned records is a technology change affecting all users in an organisation, taking them from a manual self managed style of work to an organisational systems approach with limited personal customisation. Corporate governance requirements and growing pressures for legislative compliance have stimulated public sector uptake of content management systems that can mirror internal corporate approval processes, store information and retrieve it in an accurate and timely manner (Government Exchange, 2004). These external pressures, together with the availability of funding and support for systems underpinning open and accountable government, have resulted in local government instrumentalities becoming fertile ground for full EDRMS implementations. Local government in Australia is made up of 629 Councils and 100 community governments. One of the largest of these is the City of Charles Sturt (CCS). CCS employs just over 400 full time equivalent staff in 10 business portfolios. Seven records staff members register over 300,000 records annually. In 2002 the process of replacing the Council’s paper records system began with a series of staff workshops designed to examine staff issues in moving from paper to electronic documentation. They found the paper system limited staff ability to track and share documentation; and the speed at which information could travel through the organisation. The increasing volume and use of email added to these problems. The discussion from this 56 IADIS International Conference e-Commerce 2009 series of workshops culminated in the creation of a Records Management Strategic Directions document that clearly established the need for an Electronic Document and Records Management System (EDRMS). The selection project began by establishing a project team with the responsibility ‘to replace the records section of the GEAC TCS and other localised manual and indexed systems used for the same purpose’. The initial document set out six objectives: system replacement within a set budget, a time frame for implementing the replacement system, process and efficiency gains, minimised need for new hardware/ operating systems, minimised changeover effects on running the business; and compliance with the Records Management Strategy. At this stage the size and budget of the project still had to be determined. In October 2002, the request for tender was advertised and in June 2003, contracts were signed with Tower Software for TRIM, the preferred product. Around two thirds of the CCS workforce (280 staff members) would be directly involved with TRIM usage. The implementation team consequently decided to institute an analysis of all council business processes touching records. The documentation of business process workflows at CCS began by identifying each of the 191 business units/processes; and recording/ranking interest in specific conversion processes within individual business units. The consultation process was valuable, as the rankings were helpful to management when it came to selecting ‘champions’ to promote uptake within business units. Some 10-15 of the business processes identified as conversion candidates were implemented at start-up. Vendor training had to take place on the TRIM system and be delivered as a one day mandatory session for all computer users. To ensure that all staff understood the meaning of a record, management invested in extended training time to cover Records Management procedures Groups of ten staff members were trained each day over a 7.5 week period. Three months after TRIM went live, a staff survey found staff members appreciated that so much more of their work could be done from the desktop. They were confident in the security of documents on the system and enjoyed the fact that it was now possible for multiple users to be on the system at the same time. The review also gave positive feedback on such issues as meeting the Go Live date, the high quality of training for TRIM Context, excellent change management and enhancement of the TRIM Context Workflow module through CCS staff suggestions. All records have been on TRIM since 8th March 2004. Familiarity with the EDRMS has increased staff confidence in using electronic records systems – transferable skills increasingly valued on the job market. There are still some issues with low compliance in areas such as email where decisions to ‘TRIM’ (add to records) and the time required entering each item have not yet been resolved. The success of systems integration at CCS, achieved for a project budget of $150,000, was a notable feature of the EDRMS implementation. Although not part of the original project, the TRIM Document Assembly process is now widely used for internal forms and document templates. Most importantly, the integration between TRIM, Proclaim One and GIS means that all these systems can now be viewed at desk level. For some employees the fact that there is a new baseline for business as normal continues to be a demanding concept. Organisational integration of an EWS ultimately means finding ways to maintain the impetus for cultural change (Laeven 2005). 3. THE TECHNOLOGY IMPLEMENTATION AND ACCEPTANCE MODEL (TIAM) In Section Two of this paper we concluded that IS implementation studies lack a widely accepted theoretical framework to satisfactorily explain how complex and networked technologies diffuse in practice. We found that in attempting to apply EDT and the TAM/TAM2/UTAUT approaches to the technology adoption described in the EDRMS case study, the theoretical models were either not directly applicable, or did not fulfil all these requirements. EDT was useful to address many of the consultative and expectation-setting project activities. Actions such as workshops to discover what organisational staff thought of existing paperbased processes, as well as where they wanted their organisation to be in the future, were very powerful tools for driving change. The initial support at CEO level provided leverage that helped to shape the attitude of many staff. It obliged them to try the new system for themselves and provided a “kick start” for those not truly individual innovators or early adopters in technology innovation diffusion. An initial organisational directive removed inconsistency between possible technological solutions and mandated the technology for the organisation. This explains how initial attitudes and intent were shaped in 57 ISBN: 978-972-8924-89-8 © 2009 IADIS practice. In contrast to TAM/TAM2/UTAUT, the case study shows that positive experiences after mandating the technology proved more influential than the initial mandated selection. These experiences formed a basis for building positive user support resulting in diffusion and adoption of the technology. The TAM/TAM2/UTAUT models offer a clear link between the individual’s perception of ease of use and usefulness which shapes their initial behaviour. The case study also serves to highlight the key role of timing and quality of information dissemination in ensuring perceptions are built at the right time and reinforced by practical application. In this instance, the users of the new technology were not “over-sold” on the potential benefits, but were informed of the benefits that were directly related to their originally-stated needs. While promoting additional benefits is often seen as a way of increasing an individual’s desire to adopt a new system, it can also increase perceived complexity of use. Careful and gradual release of information for any new system is vital to incremental adoption of the new technology. 3.1 Review of the Theories While both EDT and TAM/UTAUT provide insight into how a technology implementation unfolds, there are some specific limitations which are not addressed by the two theoretical frameworks. Although EDT addresses many of the factors that an organisation must deal with when facing technological change, such as understanding how choices are made between systems (and even whether to choose a system at all), the primary focus of this paper is on selecting the theory or model that can best assist following the decision to implement. We found that EDT provides an excellent explanation for why the organisation chose a particular path, whereas TAM/TAM2 and UTAUT offer a stronger theoretical framework once the decision to implement has been made. The success of an EWIS requires input across both all technology adopters, from early adopters to laggards (Rogers 1994). TAM/TAM2 and UTAUT offer significant insight into likely user reaction to the introduction of new technology. Less apparent in this model, however, is a demonstration of the feedback process between perception and action. The EDRMS case study clearly shows that technology implementation is an iterative process, with many factors shaping initial expectations, perceptions and, ultimately, attitudes to using the technology. Once individuals have used the new technology or received feedback from others, their perceptions evolve and diffuse through the organisation. This process can become a positive or negative reinforcement loop. 3.2 A Composite Theory Proposal The IS literature does not provide a widely-accepted model for EWIS uptake. Where existing theories or models may not fit the needs of a study individually, their union can provide a better explanation (Hardgrave and Johnson, 2003). Our case study findings were consistent with each of the proposed theoretical frameworks but did not offer the analytical strength of our proposed model: TIAM (Technology Implementation and Acceptance Model), where we have synthesised existing constructs and addressed limitations through the addition of other elements. We believe TIAM (see Figure 1) is a closer approximation to the reality of technology adoption and a more potent model for the introduction and diffusion of EWIS. Figure 1. Technology Implementation and Acceptance Model (TIAM) 58 IADIS International Conference e-Commerce 2009 The first panel represents one-off factors which can shape initial attitudes before the technology is experienced in a business context. Two such factors appear only in this section of the model. The initial individual subjective factors (I) are shaped by people’s impressions of: what the innovation means to them, their previous experience with technology innovation, what they think happened in other organisations; and their overall expectations of this implementation. The second factor is the initial organisational influence (O). This reflects the pressure, or expectations, placed on potential users of the technology to adopt. In general, one would hope that this is a positive pressure from a senior level of an organisation. Both I and O are very important but short term, influencing factors, which provide a “kick start” to a process that will last some time for every technological introduction. Once these factors have been used, they cannot really be reused. The CEO at Council promoted use of the new system when introduced, encouraging initially positive attitudes toward use. If the system does not deliver on these expectations, however, resistance to the system will ultimately result in its failure. The centre panel of the TIAM model represents an evolution of TAM/TAM2/UTAUT, modified to show that for organisational success it is the sum of all individuals which makes up the collective U, E, A, BI and, eventually, adoption – or, in EDT terminology, diffusion. The U and E factors are shapers of initial attitude, but are also part of a feedback loop. The sum of the entire organisation’s attitude toward use and behavioural intent is not equally spread across all users: some will have far more influence than others. For the greatest chance of success, those with the most influence should be targeted to raise their BI component as part of the organisation. This issue is especially important in the case of an EWIS, where one section of an organisation can block the diffusion of a technology innovation and thus prevent uptake by the entire organisation. The central section of the TIAM model has a feedback loop, where actual usage experience changes the U and E factors over time. If system usage fulfils or exceeds expectations, U is increased. If the technology is easier to use than expected, E will increase. The feedback loops in the TIAM model are critical to reflect reality: the two extremes of the spectrum are the first time it is used and the last time it is used before replacement or retirement. Successful diffusion lies somewhere between these two extremes and can only truly be determined by the organisation managing the technology adoption process. Organisational adoption shifts over time: what might be seen as excellent levels of use and very positive results in week 1 of a new software system is less than the adoption and usage required to sustain it long term. Similarly, if the feedback loops are not well maintained over time with new related innovation, training and positive organisational attitude, a slow drift to a more negative organisation position with less usage and lower infusion can occur. We argue that positive reinforcement is the desired result of a successful technology innovation implementation – and this takes time. If the system does not live up to expectations, reinforcement will be negative, often quite fast and preventing infusion of the technology. If failure occurs then providing sufficient new input to the cycle via inputs such as I and O is extremely difficult. The third panel of the TIAM model shows the two possible outcomes from this process. The two results are success or failure, the measures of which are subjective to the individual organisation, ranging across a wide spectrum of definitions. The figure refers to a ‘break over point’ – one at which failure of a project, once reached, is almost impossible to undo. 4. CONCLUSION AND FUTURE WORK This paper has investigated two groups of theoretical models often applied to IS research studies, to determine their value as predictors of technology implementations. Imported from the increasingly crossdisciplinary field of innovation research, Diffusion of Innovation Theory (EDT) plus the Technology Acceptance Model (TAM/ TAM2) and Unified Theory of Acceptance and Use of Technology (UTAUT) are all well accepted in IS research. Together they provide an enhanced explanatory framework for uptake of new technology. Following adaptation of key features of the two frameworks to our Technology Implementation and Acceptance Model (TIAM) we showed how TIAM can be applied to a real-world implementation of an EDRMS. We found that TIAM provides a theoretical framework highly applicable to IS practice in both private and public sector contexts, particularly for technology adoption in local councils and small to medium sized enterprise (SMEs). The TIAM composite model is structured in an easily understood and interpretable format 59 ISBN: 978-972-8924-89-8 © 2009 IADIS and is directed at use internally by the implementing organization. Practitioners can apply the model to provide increased clarity and simplicity and to enable more effective collaboration across the organisation. TIAM enables communication of key signs to parties not versed in project management and technology implementation indicating how the process is likely to unfold and why and where decisions need to be made. Although our initial findings on the applicability of TIAM were extremely favourable, we are aware that further testing is needed to verify the usefulness of the framework, to establish whether the model can change behaviour and provide useful insight to project managers enabling them to evaluate projects prior to failure. In future work, the authors therefore propose to further test and validate TIAM as a theoretical framework that can effectively underpin IS project management by applying it to a series of case studies from different industries. REFERENCES Agarwal R. and Prasad, J. (1997) “The role of innovation characteristics and perceived voluntariness in the acceptance of information technologies” Decision Sciences, Vol. 28, 3 pp.557–582. Ajzen, I. and Fishbein, M. (1980) “Understanding Attitudes and Predicting Social Behavior” Englewood Cliffs, NJ: Prentice-Hall,Inc. Arthur, W. B. (1994) “Increasing Returns and Path Dependency in the Economy” Univ of Michigan Press, Ann Arbor. Bagozzi, R. P., Davis, F. D. and Warshaw, P. R. (1992) “Development and test of a theory of technological learning and usage” Human Relations, Vol. 45, 7, pp.660-686. Burton-Jones, A. and Hubona, G.S. (2005) "Individual Differences and Usage Behavior: Revisiting a Technology Acceptance Model Assumption," The DATABASE for Advances in Information Systems, Vol 36, 2 pp. 58-77. Bijker, W. B. and Law, J. (1992) “Shaping Technology/Building Society: Studies in Sociotechnical Change” in Bijker, W,Carslon, W.B. & Pinch T. (eds.) Inside Technology MIT Press Cambridge, Mass.: MIT Press, pp.205-240. Carlsson, C., Carlsson, J., Hyvonen, K., Puhakainen, J. and Walden, P. (2006) “Adoption of Mobile Devices/Services — Searching for Answers with the UTAUT” Proceedings 39th Hawaii International Conference on System Sciences, Volume 6, 04-07, pp.132a - 132a Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989) “User acceptance of computer technology: A comparison of two theoretical models”. Management Science, Vol.35(8), pp.982-1003. Government Exchange (2004) “Electronic Documents and Records Management for the Public Sector” http://www.iqpc.co.uk/binary-data/IQPC_CONFEVENT/pdf_file/5799.pdf (retrieved Feb 28th, 2008). Hardgrave B.C. and Johnson, R.A. (2003) “Towards an information systems development acceptance model: the case of object-oriented systems development”. IEEE Transactions on Engineering Management Vol.50, 3, pp.322-336. Laeven T (2005) Competencies – the asset that counts most: on developing human talents as a prerequisite for successful EDRM changes pp 129 -148 in Hare C and McLeod J., Managing Electronic Records Facet Publishing UK. Lissoni, F. and Metcalfe, J. S. (1994), “Diffusion of Innovation Ancient and Modern: A Review of the Main Themes” in Dodgson M and Rothwell R (eds) Handbook of Industrial Innovation, Edward Elgar Publishing Limited, Cheltenham pp. 106-144. McCoy, S., Galletta, D.F. and King, W.R. (2007)"Applying TAM across cultures: the need for caution," European Journal of Information Systems Vol. 16, 1, pp. 81-90. Moore, G. C. and I. Benbasat (1991) “Development of an instrument to measure the perceptions of adopting an information technology innovation” Information Systems Research Vol 2, 3, pp 173-191. Nelson, R. R. (1995) “Recent Evolutionary Theorizing about Economic Change” Journal of Economic Literature, Vol.33, pp.48-90. Saviotti, P. P. and Metcalfe, J. S. (1991) “Evolutionary Theories of Economic and Technological Change: Present Status and Future Prospects”, Harwood Academic Publishers. Venkatesh V and Davis F D (2000) “A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies”, Management Science Vol. 46, 2, pp.186-204 Venkatesh V, Morris M G, Davis G B and Davis F D (2003) “User acceptance of information technology: toward a unified view”. MIS Quarterly 27, 3, pp.425–478 60 IADIS International Conference e-Commerce 2009 IEBS: A MODEL FOR INTELLIGENT E-BIDDING Pedro Brandão Neto, Sofiane Labidi, Rafael de Souza Cunha and Rafael Soares Cruz Intelligent System Lab. Federal University of Maranhao Campus of Bacanga 65080-040 Sao Luis –MA, Brazil ABSTRACT The interest in e-bidding systems development has increased in recent years. The e-bidding systems are focused in the crier’s role that monitors and manages the whole e-bidding process. In addition, every process is based upon laws, on an edict and on the suppliers’ database. Thus, the crier plays a main role in an e-bidding process. This paper presents a model of Intelligent E-Bidding System (IEBS) for automating all the steps of e-bidding transactions and also aims at automating the crier’s decision making role increasing, that way, the system’s autonomy. KEYWORDS Electronic Bidding; Public Administration; e-Government; Intelligent Agent. 1. INTRODUCTION With the boom of Internet, the electronic business transactions became an unavoidable trend. The ECommerce (EC) represents a significant piece of the global market and does justice to the investments made until now. If there is an area that deserves studies and efforts towards being improved, this is such area. Going through this point of view, the use of Intelligent Commerce System (ICS) and Intelligent E-Bidding System (IEBS) could represent a big profit for the improvement of web based buying & selling services. Researchers as Fonseca et al. (2003), Tomaz et al. (2004), Labidi and Martins (2007) and Cunha (2009) demonstrate the relevance of the ICS and IEBS. Statistics show that the E-Commerce in Brazil has increased exponentially (ARAÚJO, 2007). There was, in the first half of 2008, compared to 2007, a growth of 45% in revenues from companies that trade through the Internet. The total was 3.8 billion Reais, coming close to 4.4 billion Reais billed during the whole year of 2006. This phenomenon can be understood by the increase in the number of users and the expansion of the Internet all over the country. Brazil stands out among the first in number of people that use this vehicle of dissemination of information and communication (E-COMMERCE, 2008). The development of an E-Commerce system through new paradigms can affect several business areas of the companies that possibly want to adopt it. Thus, it can reduce costs of logistic, negotiation, marketing, generation of contracts and more. Thus, the need companies have to always improve and increase their capacity to interact in a more efficient way with other organizations reducing costs and increasing the profits. is very worthy. Despite the singular evolution of E-Commerce, with impressive numbers, and of its worldwide expansion that allows the breaking of boundaries, the competitiveness advancement requires that the companies search for new systems capable of increasing even more the efficiency and overcoming the technological limitations of the current software. With the emergence of more robust applications and the need to assist the human decision-making came the possibility of creating more sophisticated software systems capable of improving the business processes and to solve general scope problems. In order to create a system capable of increasing the business capacity between enterprises through a B2B relationship, the ICS (Intelligent Commerce System) aims at automating the negotiation process and the decreasing of existing costs by increasing the efficiency in relation to the current e-commerce systems, 61 ISBN: 978-972-8924-89-8 © 2009 IADIS ICS joins the characteristics of open systems such as the Internet, with an architecture that facilitates the search for trading partners, the negotiation, contracts conclusion and monitoring of contracts. All these steps are currently monitored by a human operator (TRINTA, 2007). With this in view, the ICS aims at, using the concepts of AI (Artificial Intelligence) through entities known as Software Agents and E-Commerce, creating a complete business environment for the exchange of products and services in order to provide greater efficiency in the costly process of finding business partners and also provide an reasonable way of negotiation between sellers and buyers (LABIDI at al. 2003). This paper presents a model for developing a system, called IEBS, which was designed based on the ICS in order to promote negotiations via Internet between companies and government known as E-Bidding (MEIRELLES 2005 and FÜCHTER, 2004). So, on that principle, we envisage significant company gains for businesses and government costs decrease due to automation of accomplished negotiations. Thus, it also contributes by adding the use of computers systems in the administration as an important feature for gain in business managing procedures. The extra knowledge generated by this study can contribute to researches on systems that cover other fields of administration, as well. Currently, the whole procedure realized in an E-Bidding System is intrinsically tied to laws and a legal edict. And this, respectively, dictates how the E-Bidding will be conducted according to the legal process and sometimes transcribing, fully, on its content, the text of the laws. Thus, the crier exercises a leading role in an E-Bidding. Therefore, the E-Bidding System depends on the current "presence" of the crier and his team support in all stages of a session. Therefore, the Management E-Bidding Systems today allow the crier to monitor and manage Proposals, Bids and Bidders, among other activities. Obviously, the crier acts with a team support, but the final decision comes from him. Anyway, in all stages of the competition the crier will be there to intervene when necessary. It is important to notice the differences between Proposals and Bids. The first one is the preparative stage in which Bidders make shots containing the value of their services or products. The second one is a competitive stage, after the Proposal Stage, in which Bidders make successive bids competing at lowest prices with other Bidders. Notice that the crier, even with a Management E-Bidding Systems is still the "master" and, therefore, has several tasks during the E-Bidding advancement. The higher the number of Bidders the more overloaded is the crier. This work will develop a model of Intelligent E-Bidding System (IEBS) in order to automate the EBidding System and, especially the role of decision-maker exercised by the crier, providing, therefore, greater autonomy to the E-Bidding to operate more efficiently and speedily, and with greater transparency in the process. The IEBS is based on techniques of Artificial Intelligence, in the paradigm of intelligent agents, i.e., it is oriented to intelligent agents. The main intelligent agents of IEBS are Crier Agent and Bidder Agent. These agents have a knowledge base containing the rules (knowledge of the Crier and Bidder) for the implementation of an E-Bidding. 2. INTELLIGENT E-BIDDING SYSTEM (IEBS) This section depicts the model of the Intelligent E-Bidding System (IEBS), which was developed to automate the steps of the current model of E-Bidding Systems. As a summary, the execution of an E-Bidding involves the task of a Crier that follows all the stages (or life cycle) of the E-Bidding, where the Human Crier needs to observe, examine and manually judge all Proposals and Bids offered by the Bidder (the user that represents the company). All knowledge of the steps required in the conducting of an E-Bidding is based on laws of this type of Bidding. Essentially, the edict provides rules and conditions for Bidders participation, on how to make Proposals and Bids - what is in the laws - judgment of proposals, qualification, attachments (with objective description of the objects of Bidding, and the spreadsheet cost when required), statements required to participate in the process and the term of reference. The IEBS is a new proposed model of E-Bidding based on a Multi-Agent Society that performs the goals of the system. The overall goals of the system were divided into stages that represent the specific objectives of the agents. The Intelligent Agents, that are leading the steps of proposals and bids for an E-Bidding, use 62 IADIS International Conference e-Commerce 2009 their own knowledge stored in their knowledge bases to achieve their goals. So, with the development of this new model, the E-Bidding System may have greater autonomy, without any human intervention, except during the feeding of the knowledge bases, i.e., the transfer of knowledge of how to conduct a session for the Crier and Bidder Artificial Agents. However, modeling the knowledge needed to perform an E-Bidding is not trivial. One way to qualify Crier Agent and Bidder Agent with the knowledge to conduct an E-Bidding is to extract information from the edict carefully and also get the knowledge of an E-Bidding human expert. Below is detailed the Multi-Agent Society that makes up the IEBS. 2.1 Multi-Agent Society of IEBS The Society of Intelligent Agents to perform the steps of IEBS is composed of three types of agents, two cognitive (Crier and Bidder) and a reactive (Knowledge Acquisition Agent). Each intelligent agent of the IEBS fulfills his work through the definition of Behavior. Below are defined the roles of the following agents: Bidder, Crier and Acquisition of Knowledge. • The Bidder Agent (BA) represents companies that participate in the Electronic Bidding. Its tasks are: 9 Request registration to the Crier in each Bidding. 9 Submit Proposals with reference values for the items it want to compete. 9 Submit Bids during the public session. • The Crier Agent (CA) represents the Crier (government) in the negotiation, in order to: 9 Manage the registrations in the Bidding, keeping record of all agents interested in participating of the Bidding. 9 Receive and store the Proposals of each Bidder. 9 Receive, analyze and store, in the shared database, the bids offered by agents Bidders. • Announce and formalize, at the end of the public Bidding session, the official winner. • The Knowledge Acquisition Agent (KAA): 9 It will be constantly waiting for messages from the environment with the intermediation of the jadeGateway agent. The content of the message are the rules and standards of the session that was provided by the crier or bidder. 9 It will convert the contents of a message, received from the Gateway Agent, to knowledge in the JESS and then send a message representing the knowledge in JESS to the Crier Agent or Bidder Agent. For the IEBS modeling, the following use cases were defined in the topics bellow: 2.1.1 Opening of the Edict – Feeding of the Knowledge Base In the current model of E-Bidding the composition and elaboration of the edict is essential. This is because it is a document that formalizes and sets rules and standards of an E-Bidding and describes the object that will be negotiated. However, in this new model of the E-bidding, the edict needs special care in its preparation, since in the IEBS, the intelligent agents rely on knowledge established in the edict for the E-Bidding conduction. After approval of the edict, the Crier and his team provide to the system the rules (knowledge) of how the session will be conducted. Subsequently, the Crier Agent and the Knowledge Acquisition Agent are instantiated. So, declared the opening of edict, the Bidders Companies interested to participate in the EBidding can already sign up. As explained earlier, the edict contains all the knowledge required to perform an E-Bidding. Therefore, for each Crier Agent, there will be knowledge base "supply". The definition and opening of the edict is one of the most important use cases of IEBS, representing the first stage of the life cycle of E-Bidding. This use case is also the only human intervention (referring to the Crier Agent) during the process. After the implementation there is a Crier Agent for that session, which will lead the whole process of autonomous E-Bidding. It is important to note that all interaction between users and the system is mediated by KAA. This agent captures the data provided by the user in the graphical interface, through an intermediary agent and inserts 63 ISBN: 978-972-8924-89-8 © 2009 IADIS them in the knowledge base of the Crier Agent. Therefore, there is a need for two interfaces agents, one into which the user feeds the knowledge base and the other is the interface mediator agent. 2.1.2 E-Bidding Registration In the moment of system execution the Bidder Agents are instantiated and associated/registered in a session. Its knowledge base is "fueled" by the representatives of the company and will be composed of the necessary knowledge to help the agent to decide (or assist in the decision of the company) the best Bid to be offered. After the confirmation of registration in the E-Bidding, Bidders Agents are instantiated in the system and belong exclusively to the E-Bidding they have been registered to. In the next stage of IEBS implementation, when the whole process of E-Bidding (since the edict until the contract) is conducted by autonomous agents with minimal human intervention, the process of associating the Bidder Agents to the specific E-Bidding is done automatically: the own, driven by corporate interests they represent, require registration in the Bidding. Such E-Biddings, when registered in the system, will be available through a service similar to yellow pages, to all the agents who compose the society. It is also characterized as an auxiliary use case for the IEBS (Proposals and Bids Evaluation) module projected to be implemented, once in the complete system, the stage of registration of Bidder Agent will be given at the time of the company's registration in the system, and the inclusion of this agent in any E-Bidding occurs autonomously and at different times of the process. This is not the only human intervention on the performance of this agent, as its base of knowledge will possibly update (according to the changes in the interests of the company). 2.1.3 Proposals Making Is the use case that represents the second moment of execution of the agents (the first time the Bidder Agents are registered in the E-Bidding) in which Bidder Agents (following the Proposals Negotiation Protocol to be explained next) send messages containing the data of its Proposal to the Crier Agent. Therefore, the Proposals aim at legitimizing the E-Bidding process and encouraging lower prices from the beginning, increasing competition. Thus, the intention is actually preparing the framework for the main part which is the stage of Bids. After the Bidder’s registration in the E-Bidding, the initial Proposals related to E-Bidding items the agent wishes to participate in are sent by the own. 2.1.4 Proposals Evaluation The Crier Agent receives the proposals sent by the participants, then, analyzes (or judges) its acceptability. To review the acceptability of the Proposals, the Crier Agent consults its knowledge base in search of the trial requirements stipulated in the edict. 2.1.5 Bids Making Is the use case that emerges in the third time agents execution (third step of the Bids and Proposals Evaluation Stage), which contains the main interaction between agents. At this stage the Bidder Agents may offer Bid for a period of time, previously specified in the edict. After this period there is a winner of the session (the one who proposed the lowest bid). All bids are evaluated by the Crier Agent based on criteria defined in the law or edict. The following subitem describes such use case. 2.1.6 Bids Evaluation The Crier Agent will manage the Bids and make available to all, through a shared database, information about the offered Bids: Bid and reference to the Bidder that offered it. The reference is a method of ensuring that the process occurs in a transparent way, in which the name of the company is not disclosed until the event is finished. Finally, the Crier Agent officially announces the winner of the event. 3. IEBS PROTOTYPE This chapter describes the IEBS implementation, emphasizing how the Intelligent Agents operate in EBidding steps. Thus, in the agents development the JADE (Java Agent Development Framework) platform 64 IADIS International Conference e-Commerce 2009 (JADE, 2008) and the Jess inference engine (Java Expert System Shell) (JESS, 2008) were used. JADE provides ready libraries, the environment, implementation and graphical tools for managing the agent society; thus facilitating the development of these agents. The JESS enables the agents with the capacity to infer, because it defines a set of rules that act on a Knowledge Base. Intelligent Agents operate in IEBS using negotiation protocols. The architecture of the IEBS, as shown in Figure 1, is composed by the Crier Agent (CA), Bidder Agent (BA) - with their knowledge bases - and Knowledge Acquisition Agent (KAA), which appears as an auxiliary agent, being responsible for the construction knowledge base of the Crier Agents and Bidder Agents, capturing the knowledge provided via web interface. In this society there is a shared database (with information about the Bids offered in the edict), to which all agents will have access (will be able to read), but only the Crier Agent will be able to edit. Figure 1. IEBS Architecture The Crier and the Bidder have a web interface to supply the knowledge base. It was necessary to use a Proxy Agent, here called JadeGateway (agent whose implementation is provided with the JADE framework), to establish communication between the external environment (web interface) and the Agents Society. This integration was necessary to establish communication with the Agent Society, because IEBS is a Web-Based E-Bidding System. Then, the data provided in the web interface are captured by the Proxy Agent who then sends to KAA. It gets the message and converts the information into JESS rules and encapsulates them into an object and finally sends it to the Crier Agent or Bidder Agent that inserts it in their knowledge base invoking its method. The Database (DB) is a table of the system that is shared by all users. Its goal is to record every Bid. In this prototype, project details of foreign agent systems, such as the user interfaces, integration with other systems, are separated from its implementation, focusing to limit the IEBS implementation to the agents operations and interactions, etc. 3.1 Interaction Protocols between the Agents IEBS agents will exchange messages to achieve their goals. The negotiation protocol will be used. In this case, the problem of tasks assignment will be solved by induction of the coordinator and the actors to follow a sequence of messages, where the initiator is the coordinator and the participant is the actor. This message sequence is known as a negotiation protocol. Several common situations can be conducted by the adoption of pattern interaction protocols, such as auction, registration to receive notifications, negotiations and so on. Such protocols are specified by FIPA (Foundation for Intelligent Physical Agents) and JADE (Java Agent Development Framework, used in the project of Multi-Agent Society) implements them offering classes that easy the programmer’s work by reducing the overload caused by the need for checking the flow of messages when two or more agents interact following an interaction protocol. These classes provide return methods that should be redefined by the programmer through the insertion of logic associated to the specific domain. 65 ISBN: 978-972-8924-89-8 © 2009 IADIS 3.2 IEBS Utilized Protocols Among the protocols specified by FIPA and implemented by JADE, three are compatible and adaptable Among the protocols specified by FIPA and implemented by JADE, three are compatible and adaptable to the reality of E-Bidding and the implementation of Proposals and Bids Evaluation Stage. They are called adaptive because its methods can be overwritten with the required logic. These methods are implemented according to the occurrence of events (events like the arrival, deadline timeout etc). The protocols are the following: 3.2.1 Registration Protocols Adapted from FIPA-Request protocol (implemented in the AchieveREInitiator and ArchieveREResponder classes). In this protocol, the interaction between actors is initiated by the Bidder Agent, responsible for sending messages to the Crier Agent, telling of his participation in the E-Bidding. Another message is returned by the Crier Agent informing the refusal of registration and its justification. The absence of a return message is treated as success in the application. A new interaction between the agents is started in case the Bidder Agent wants to try again to inform the Crier Agent about his registration. More details about the flow of messages are given in the figure 2 below: Figure 2. Messages Flow 3.2.2 Proposal Protocol Adapted from protocol FIPA-Propose (ProposeInitiator and ProposeResponder classes implemented in the framework JADE). For the Proposals Evaluation was implemented the Proposals Protocol, used by the Bidder Agent which is responsible for the interaction initiation, by sending message with details of the Proposal to the Crier Agent. In response, the Crier Agent can return a message denying the Proposal (and the reasons for refusal) or not return any message, which indicates the acceptance of the Proposal, as seen in the figure 3 below: 66 IADIS International Conference e-Commerce 2009 Figure 3. Messages Flow 3.2.3 Bid Protocol Adapted from FIPA-Contract-Net protocol (implemented in the ContractNetInitiator and ContractNetResponder classes, present in the JADE framework). In the third moment of agent performance, in which occurs the transmission and evaluation of Bids and when the Bid Protocol is used, the Crier Agent is responsible for initiating the protocol by sending messages to all Bidders Agents able to participate in public session. These messages are of CFP (call for Proposals or request for Proposals) type and indicate the opening of E-Bidding Bids. The following messages from the Bidder Agents, still in the same protocol, will be composed by Bids offered by each one. When receiving the Bids, the Crier Agent evaluates and returns a message in case of refusal (justifying it) or returns none, what will be understood as the acceptance of Bid. More details are given in figure 4 below: Figure 4. Messages Flow 4. CONCLUSION In the current E-Bidding Systems panorama there is still overload in Human Crier work . Therefore, the main contribution of this work was the development of a model called IEBS system as to automate the stages of the E-Bidding. The IEBS has the following advantages: i) simple architecture, easy understanding and implementation, ii) knowledge of the Crier will be separated and represented with production rules in its Knowledge Base, iii) ability to infer; iv) interaction protocols were defined for the Proposals and Bids stage. Therefore, this system will facilitate the task of the Crier in conducting and monitoring all E-Bidding and thus reduce the Crier’s overload of activities. 67 ISBN: 978-972-8924-89-8 © 2009 IADIS It is extremely difficult and delicate to define the Knowledge Base of the Human Crier because there is no standard model of edict to be followed. However, the knowledge for the accomplishment of a session is tied to the edict and laws. The IEBS can still be improved working specifically on the Knowledge Base of Crier Agents and Bidder Agents. In the future, it will be possible to turn them more independent so that they may lead all stages of the process, even independently of the entities they represent. REFERENCES ARAÚJO, Antonio da Silva. 2007. Economia das Compras governamentais em decorrência do pregão Eletrônico - uma abordagem econométrica. Dissertação (Mestrado em Economia) – Universidade Federal do Ceará. Fortaleza. BECK, K. and Ralph, J. 1994. Patterns Generates Architectures. Proceedings of European Conference of ObjectOriented Programming. Bologna, Italy, pp. 139-149. CUNHA, Rafael S. 2009. Proposta de Desenvolvimento de um Sistema Inteligente de Pregão Eletrônico Baseado no Sistema ICS de Comércio Eletrônico: Uma abordagem descritiva. Monografia - Universidade Estadual do Maranhão, São luis, Brasil. ESTATÍSTICAS DE COMÉRCIO ELETRÔNICO NO BRASIL. Disponível em: http://e-commerce.org.br/stats.php Acesso em: 23/12/2008. FONSECA, Luis Carlos et al. 2003 ICS – An Agent Mediated E-Commerce System: Ontology Usage. In: International Conference on Enterprise Information Systems. Angers. Proceedings of the 5th International Conference on Enterprise Information Systems. FONSECA, Luís Carlos Costa. 2003. Sistemas Multiagentes para Negociação no Ambiente ICS de Comércio Eletrônico. Dissertação (Mestrado em Ciência da Computação) – Curso de Pós-Graduação em Engenharia de Eletricidade, Universidade Federal do Maranhão, São Luís, Brasil. JADE – Java Agent Development Framework. Disponível em: < http://jade.tilab.com/ >. Acessado em 14 nov. 2009. JESS – Java Expert Shell System. Disponível em: < http://jade.tilab.com/ >. Acessado em 6 out. 2009. LABIDI, Sofiane et al. 2007. Negotiation Model of the ICS Enviroment. In: Third International Workshop on Data Engineering Issues in E-Commerce and Services, 2007, San Diego, CA. DEECS. MEIRELLES, Hely Lopes. 2005. Direito Administrativo Brasileir., 30. ed. São Paulo: Malheiros Editores. NOLETO, Walter D. C. 2007. Pregão Presencial e Eletrônico: Análise do Instituto com enfoque em alguns Pontos Polêmicos. Monografia – Curso de Pós-graduação Lato Sensu em Direito Uniplac. Brasilia. FÜCHTER, Simone Keller. 2004. Adaptive Fuzzy Approach to Estimate Supplier’s Competitiveness in Open e-Bidding. In: IEEE International Conference on Ecommerce, Beijing, China. TOMAZ, Ricardo et al. 2004. Composition of Web Services in the ICS Architecture. In: 6 th International Conference on Enterprise Information Systems, Porto. ICEIS-2004 Proceedings. TRINTA, Valeska Rogéria Vieira. 2007. Modelagem do Agente de Contrato Eletrônico da Fase de Formação de Contratos no Sistema de Comércio Inteligente (ICS) Considerando a Lei Modelo sobre Comércio Eletrônico. Dissertação (Mestrado em Ciência da Computação) – Curso de Pós-Graduação em Engenharia de Eletricidade, Universidade Federal do Maranhão, São Luis, Brasil. 68 IADIS International Conference e-Commerce 2009 AN APROACH FOR SEMANTIC-BASED EC MIDDLEWARE Ejub Kajan High School of Applied Studies Filipa Filipovica 20, Vranje, Serbia Leonid Stoimenov Faculty of Electronic Engineering Beogradska 14, Nis, Serbia ABSTRACT In this paper we are introducing an advanced architectural framework for open interactions between e-business applications. After short analysis of the requirements there is an overview of the related work elsewhere. Finally, our approach, named as B2BOOM, is described. Concluding remarks emphasize the importance of such semantic based approach and give future directions. KEYWORDS Electronic commerce, interoperability, middleware, ontologies. 1. INTRODUCTION The idea to build a distributed open network capable of exchanging business data was born in the middle 1980s. Despite good reasons, this initiative, known as EDI (Electronic Data Interchange), failed due to its robust specification, expensive implementation and therefore limited number of participants. Next attempt, known as dotcom era, has also failed because too many companies stayed out again (e.g. most of SMEs). After disillusionment and failure analysis, new expansion of EC (Electronic Commerce) is taking place, especially in the form of B2B (Business-to Business). There are huge number of difficulties and barriers in achieving full potential of EC. Many of them are caused by a big heterogeneity between business processes, supported applications and associated data at one side and different hardware, operating systems, database systems, network infrastructure, etc. at the other side (Ng et al., 2000; Dogac and Cingil, 2001; Omelayenko and Fensel, 2001; Medjahed et al., 2003; Kajan, 2004; Kajan and Stoimenov, 2005; Hepp, 2006). This heterogeneity is the source of syntactic and semantic conflicts at all interoperability layers (Stoimenov et al., 2006) and especially on data and process layers (Kajan 2004). The general approach for solving heterogeneity problems is to define a common template, a framework, here a B2B framework capable to solve B2B heterogeneity in order to achieve business processes communication on demand despite of applications and technologies used by business entities (here and after be). This paper deals with semantic conflicts in B2B communications and it is organized as follows. The second section analyses common and specific causes of semantic conflicts. The section three is dedicated to related work in the area of solving these conflicts. Section four gives a general overview of the proposed B2BOOM middleware architecture and detailed description of approach applied to solve semantic heterogeneity on the data and process layer of business process communications. 69 ISBN: 978-972-8924-89-8 © 2009 IADIS 2. CONFLICTS ARE EVERYWHERE If two business entities are going to establish a business relationship on demand they will be faced with one or more (that is more likely to occur) of the following possible conflicts: • Data conflicts - in data itself and/or data schemas, • Document conflicts - in document names, organization of their contents, etc, • Conflicts between their business protocols – in different structure, timing and scenario of a protocol. Data conflicts are not particularly specific for EC field only. They appear as the main obstacle in any area where interoperability between two peers is required and they are on the research focus for years (Kim and Seo, 1991; Shet and Kashyap, 1992; Goh et al, 1999; Wache et al, 2001). At the data level, the different representation and interpretation of the same or similar data (e.g. expressions, units and precision) are very common (Kim and Seo, 1991). At the schema level various conflicts such those between names, entity identifiers, schema isomorphism, or many other schema discrepancies may be found (Sheth and Kashyap, 1992). Two types of naming conflicts are widely recognized, synonyms and homonyms, both coming from inconsistent assigning names to schema elements. When the same concepts such are entity relations or their attributes are described by different names it goes to appearance of synonyms. For example, if there are few business entities whose databases use different names for clients, e.g. customer, buyer, consumer, etc. the data interoperability between each of them may be very difficult, and may lead to the “Tower of Babel”. Homonyms appear when two schemas use the same name for two different concepts. For example one database may use the name “Name” for business partners, whilst in the other database it is assigned for merchandise. Entity identifiers cause conflicts when different primary keys are assigned to the same concepts in different data bases. For example, one data base may use an unique tax identifier as a primary key for a “Buyer”, whilst the other data base it may be organized as an attribute pair (e.g., Buyer name, address). When the same concepts are described by different set of attributes they arise to schema isomorphism conflicts. When the same real object has represented by two different concepts in two schemas then type conflicts appear, e.g. a bank account may be assigned in one schema as an entity type, whilst in another schema it may appear as attribute. Cardinality conflicts arise by different rules used by different business entities, thus the relationships between objects may have different extreme values. The common semantic conflicts described above appear in business data also, but there are also specific conflicts that may occur between business data. There are several reasons for this including, but not limited to, different document names, different document structure, document elements and their representation, etc. (Guo, 2006). But these conflicts are not yield to the end of the interoperability problems. As we moving to the upper layer of a business communication where business processes try to establish a business scenario, the number of conflicts is unpredictable. At this layer there is not standardized business terminology, there are not common acceptable grammar as well as dialog protocols. An analytical research (Zhao, 2005) gives an overview of such heterogeneity between four frameworks on a simple business process “Order”. This simple example shows too much discrepancy between these “orders” in their atomic transactions, including but not limited to their names, number, sequencing, etc. In reality, there are more then four frameworks in use and much more business processes. These differences make business entities unable to establish collaboration between their business processes. 3. RELATED WORK On the way from a single computer to a distributed global network, computers were going to be more and more loose and distant. With the growth of loosely-coupled computers, the desire for data integration as well as for process interoperability has also grown. At the same time the required semantic that may satisfy integration requirements has getting stronger, as shown in Table 1. 70 IADIS International Conference e-Commerce 2009 Table 1. Relationships between required semantics and coupling with enabling technologies and frameworks. Computing environment Single system Coupling Very high Required semantics Very low Enabling technologies Any commercial DBMS or file system Same DBMS LAN High Low Business entity Medium-low Medium-high FDBMS, DW, Datamarts, RPC, XML, WS,… B2B community Low-very low High-very high XML, WS,… The whole Internet Extremely low Extremely high XML, WS, RDF, RDFS, OWL, OWL-S, agents, future Intelligent Web technologies Frameworks Appropriate for a system Appropriate for LAN RPC-based, CORBA, EJB, DCOM, MOM, Workflows, .WSbased. Not yet completely solved ebXML, RosettaNet, Integrated vendor solutions (.NET, SuneOne, etc…) Not yet completely solved SESA Under development At the first sight it looks like that some standard document and data format may be used to solve these interoperability problems. During the years many middleware technologies and standard frameworks were developed in order to meet interoperability inside business entities and between them, as shown in table 1. An overview of these technologies, their strengths and weaknesses, and comparative analysis can be found elsewhere (Dogac and Cingil, 2001; Medjahed et al, 2003; Kajan and Stoimenov, 2005). Table 1 shows which of these technologies and frameworks may be used to solve interoperability requirements in different computing environments (in terms of coupling and semantic). In the meantime many “standardized” product catalogs have been proposed and deployed (Schmitz et al, 2005). These standards make additional interoperability problems as well as many classification schemas in use (Hepp et al, 2005). At the process layer situation is similar, where more than thirty reference process models exist (Fettke, et al, 2005), resulting with a number of mutually incompatible “Orders” as described in previous section. Business processes interoperability is one of the hot research topics today. It is focused around two axes; one is based on Web services (Austin et al, 2004) and the other on the Semantic Web technologies. These include analysis of potential of service technologies (Kreger, 2003; Petrie and Bussler, 2003; Kajan, 2004; Bui and Gachet, 2005; Kajan and Stoimenov, 2005; Papazoglou and den Heuvel, 2007), semantically enabled service-oriented architectures (SESA) (Patil et al, 2004; Haselwanter et al, 2006; Shafiq et al, 2007; Vitvar et al, 2007), and modeling of choreography components for business processes harmonization according to their structure and timing (Papazoglou, 2003; Staab, 2003; Cimpian and Mocan 2005; Svirkas et al, 2006; Ye et al, 2006). There are also several research prototypes of ontology-based general-purpose middleware. Examples include, but not limited to: BUSTER (Viser, 2004), CREAM (Park and Ram, 2004), OBSERVER (Mena et al, 2000), ORHIDEA (Stoimenov et al, 2006), etc. None of them do address either Web services or the problems of process choreography. The most completed effort to define an ontology-based middleware which takes care about above issues is WSMX (W3C, 2005). WSMX (Web Services Modeling Execution Environment) allows discovering, selection, mediation, invocation and mutual work of semantic Web services. The WSMX consists of a number of pairs (wrapper/interface) each of which is specialized for a specific task. These tasks include communications, resources, discovering and selection of Web services, data mediation, process mediation, and finally Web services choreography and orchestration. In order to solve semantic conflicts mentioned in section two, WSMX uses four types of mediators (OO, GG, WG, and WW), where O, G and W stands for ontology, goals and Web services, respectively. 71 ISBN: 978-972-8924-89-8 © 2009 IADIS 4. B2BOOM FRAMEWORK This section briefly described the B2BOOM (Business-to-Business Ontology-Oriented Middleware) architecture that is based on semantic brokers that use hybrid mediation in order to solve semantic conflicts between business data and processes. 4.1 Architectural Overview B2BOOM consists of several loosely-coupled software components with well-defined interfaces, the heart of which is pair of semantic brokers (data and process mediators), as shown in Fig.1. Figure1. The simplified view of B2BOOM architecture Table 2. Core components of B2BOOM framework. B2BOOM component Business entities be Domain-oriented data sources B2B ProcessWrappers B2B ProcessBrokers B2B DataWrappers B2B DataBrokers B2B Shared server Role B2B applications that belong to autonomous companies and or applications that act inside a B2B community. Business databases or their parts publicly available inside the community. They include both, data and public parts of business processes. Transform local data about processes and their schemas into the process model used by ProcessBrokers. Semantic mediators used to integrate a number of independent processes bridging semantic differences among them and to arrange appropriate choreography between their atomic public transactions. Transform local data and their schemas into the data model used by DataBrokers. Semantic data mediators used to integrate a number of independent information sources bridging semantic differences among them. Common server that consist meta-knowledge of the B2B community and that also serves as security and administration server for the community. The core components of B2BOOM mediation at the content layer are DataBrokers. Basically, B2B DataBroker is an ORHIDEA mediator (Stoimenov et al, 2006) extended with P2P connection with relevant process broker and by WSMO (a part of WSMX repository) listener. Instead of relevant data broker, the relevant B2B ProcessBroker is responsible for B2B DataBroker initialization. Therefore, the broker register 72 IADIS International Conference e-Commerce 2009 holds data about B2B ProcessBrokers in a community. There is ability to update this register statically or dynamically. DBM (DataBroker Manager) is a key component responsible for all communications inside the Broker and with relevant ProcessBrokers and DataWrappers. Using the Data Translator, DBM fetches data from B2B ProcessBroker and calls the Wrapper Finder to find which of the DataWrappers is responsible for local data source. The Wrapper Register is a table consisting of data about available DataWrappers, their identification and access mechanisms. It is loaded and updated either by system administrator or, on case of that Data Wrappers are publicized as Web services, via UDDI. Data Translator is responsible for communication with B2B ProcessBrokers. It fetches semantically conflict data from ProcessBrokers and after semantic processing returns these data to relevant business process via the same ProcessBroker. Information about ProcessBrokers in a B2B community (identification, address, access protocols, etc.) is hold by the ProcessBroker Register. This register, initially empty, is dynamically loaded by events. Data about all B2B ProcessBrokers in a B2B community is available in a special UDDI register. Semantic conflicts are solving by Rule Manger and Ontology Manager. Both components use the Knowledge Base consisting of dictionary, thesaurus, and semantic rules as well as metadata about reference model. Rule Manager communicates with Ontology Manager and Knowledge Base and also has a decision mechanism used for rules interpretation from Knowledge Base. In case that rules are not precisely as required or in case that they are not obvious, Rule Manager have an access mechanism for the domain expert who has the possibility to fine tune the rules or, if necessary, to add new rules. B2B Data Manager correlates between ontologies and reference model. It allows access to metadata and makes mappings between concepts from ontologies and classes from reference model. Finally, WSMO listener is listening to WSMO objects via WSMX resource manager (RM) component, and in relevance to its DataBroker needs, fetches the links to any of the required WSMO objects regardless that are Web services, goals, ontologies or mediators. And vice versa, WSMX RM may register into WSMO mediators a B2BOOM DataBroker. In that way B2BOOM may be an autonomous part of some WSMX cluster. The architecture of the semantic part of B2B ProcessBroker is similar. The main difference is the reference process model instead of data model. Choreography module of the B2B ProcessBroker is strictly limited to six possible scenarios as follows: (0) there are not choreography conflicts, skip it; (1) divide process into two sub-processes; (2) concatenate two processes into one; (3) exchange the order between two; (4) cancel one of the processes for a moment; (5) mediation impossible, cancel all, reset to initial state. Choreographies of all public processes are stored in a dedicated registry on shared server together with conversation id that is used to assign business entities with their public processes. That id is later used by choreography module to choose the most appropriate scenario, as mentioned before, for a given conversation. 4.2 Solving Semantic Conflicts Hybrid ontology approach (see Fig. 2a) uses three types of semantic mappings between concepts from two local ontologies: (1) direct semantic relationships between two ontologies: (2) indirect semantic relationships across top-level ontology; and (3) semantic mappings across reference common model. At the process layer, at the moment, step 3 has skipped, because we have not defined any appropriate reference ontology yet. At the data layer we used BMECat (Schmitz et al, 2004) as the reference ontology, a part of which is shown in Fig. 2b. Such approach implicates several prerequisites for the functionality of B2B community, especially that local data ontologies and local process ontologies for every process used by business entities should exist. If business entities are not bring them the B2B shared server engages the relevant broker to find missing ontologies somewhere else (e.g. in the WSMX repository of ontologies). The second request is an agreement between business entities that are participating in the community on shared components, data and public processes. 73 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 2a. Hybrid ontology approach Figure 2b. A part of BMEcat reference ontology Semantic mediation is taking place as shown in Fig. 3. All coordination is upon the ProcessBroker. It takes ontological instance from ben, extracts relevant data and sends them, if necessary, to DataBroker for semantic alignment. After data are semantically adjusted, ProcessBroker solves semantic conflicts between process data and makes necessary choreography steps (as explained in section 4.1) for particular ontological instance. Such a “micro-message” is then sending to bek together with the address of ben, that is retransmit it to ben. The loop is repeated until all ontological instances residing in the public process do not pass. If any of the mediators failed on any step, the error message is generated, sent to both business entities, communication is ending, while business processes in both entities are going to be reset in the initial state. Figure 3. Semantic mediation in B2BOOM 74 IADIS International Conference e-Commerce 2009 5. CONCLUSION The goal of the proposed open architecture is to develop reference architecture for any B2B community and any business entity for use, refine and later implement regardless of any kind of business, business model or number of participants. The concept shown is not completely tested in real environment. The core components of data mediation have been implemented and deployed; the others are still under development and must be tested in both, experimental and real environments. At the moment three important components are under investigation, the reference process model, user interface for domain expert, and choreography module testing. There are also the other milestone activities that should be done in order to fully deployment of such complex architecture. These include the development and deployment of the security plane. ACKNOWLEDGEMENT A part of this research has been supported by the Ministry of Science of the Republic of Serbia. REFERENCES Austin, D. et al. (Eds). 2004. Web Services Architecture Requirements. http://www.w3.org/TR/2004/NOTE-wsa-reqs20040211 Bui T. and Gachet A. 2005. Web services for negotiation and bargaining in electronic markets: Design requirements and implementation framework. In Proceedings of 38th HICSS, Hawaii. Cimpian E. and Mocan A. WSMX Process Mediation Based on Choreographies, 1st Int’l WS on Web Services Choreography and Orchestration for BPM, Nancy, France, Dogac A. and Cingil I. 2001. A survey and comparison of business-to-business E-commerce frameworks“, ACM SIGecom Exchanges, vol. 2, no. 2, pp. 16-27. Fettke P. et al. 2005. Business Process Reference Models: Survey and Classification. Proceedings of the BPRM Workshop, Nancy, France, pp. 469–483. Goh C. et al, 1999. Context Interchange: New Features and Formalism for Intelligent Information Integration. ACM Transactions of Information Systems. Vol. 17, No. 3, pp. 270–293. Haselwanter T. et al. 2006. Dynamic B2B Integration on the Semantic Web Services. In SWS Challenge Phase 2, Budva, Montenegro. Hepp M. et al, 2005. A Quantitative Analysis of ECl@ss, UNSPSC, eOTD, and RNTD: Content, Coverage and Maintenance. Proceedings of IEEE ICEBE, Beijing, China, pp. 527–581. Hepp M. 2006. The True Complexity of Product Representation in the Semantic Web. Proceedings of ECIS’06, Gothenburg, Sweden. Kajan E. 2004. The Maturity of Open Systems for B2B“, ACM SIGecom Exchanges, vol. 5, no. 2, 2004, pp. 34-44. Kajan E. Stoimenov L. Toward an ontology-driven architectural framework for B2B. Communications of the ACM, vol. 48, No. 12, 2005, pp. 60-66. Kim W. and Seo J. 1991. Classifying Schematic and Data Heterogeneity in Multidatabase Systems. , IEEE Computer, vol. 24, No. 12, pp. 12-18. Kreger H. 2003. Fulfilling the Web Services Promise. Communication of the ACM, Vol. 48, No 6, pp. 29–34. Medjahed B. et al. 2003.Business-to-business interactions and enabling technologies, The VLDB journal, Vol. 12, pp. 5985. Mena E. et al. 2000. OBSERVER: An Approach for Query Processing in Global Information Systems based on Interoperation Across Pre-existing Ontologies. Distributed and Parallel Databases, Vol. 8, No. 2, pp. 223-272. Omelayenko B. and Fensel D. 2001. An Analysis of B2B Catalogue Integration Problem: Content and Document Integration, Proceedings of ICEIS’01, Setubal, Portugal. Papazoglou M. 2003. Web Services and Business Transactions. World Wide Web: Internet and Web Information Systems. 6, pp. 49–91. Papazoglou M. and den Heuvel W-J. 2007. Service oriented architectures: approaches, technologies and research issues. The VLDB Journal, 16, pp. 389–415. 75 ISBN: 978-972-8924-89-8 © 2009 IADIS Park J and Ram S. 2004. Information Systems Interoperability: What lies beneath? ACM Transactions on Information Systems, Vol. 22, No. 4. pp. 9-21. Patil A., Oundakar S., Sheth A. and Verma K. 2004. METEOR-S Web Service Annotation Framework. Proceeding of ACM WWW’2004. New York, NY, pp. 553–562. Petrie C. and Bussler C. 2003. Service Agents and Virtual Enterprises: A survey. IEEE Intelligent Systems, July/August 2003, pp. 68-78. Schmitz, V, et al. 2004. XML-Based Data Exchange of Product Model in E-Procumerent and E-Sales: The case of BMEcat 2.0. Proceedings of PETE04, Copenhagen, Denmark, pp. 97-107. Schmitz V. et al. 2005 Do e-catalog standards support advanced processes in B2B e-commerce? Findings from the CEN/ISSS workshop eCAT. Proceedings of the 38th HICSS, Hawaii, pp. 1–10. Shafiq O. Zaremba M. and Fensel D. 2007. On communication and coordination issues of Semantic Web Services. Proceedings of ICWIS 2007. Sheth A. and Kashyap V. 1992. So far (schematically), yet so near (semantically). Proceedings of IFIP WG2.6 Database Semantics Conference on Interoperable Database Systems, Lorne, Australia, pp. 283-312. Staab S. 2003. Web Services: Been There, Done That? IEEE Intelligent Systems. January/February 2003. pp. 72–85. Stoimenov L., Kajan E. and Djordjevic-Kajan S. 2006. Ontology-Driven Semantic Interoperability. International Review of Computers and Software, Vol. 1, No. 2. pp. 132-136. Svirkas A., Roberts B. and Vilson M. 2006. Peer-to-Peer Business Communities: Mapping Public Coreography Protocols to Individual Implementation Mechanisms. Proceedings of IADIS E-commerce, Barcelona, Spain, pp. 226-233. Visser U. 2004. General Approach of Buster, LNCS 3159, Springer-Verlag. Vitvar T. et al. 2007. Semantic SOA to Promote Integration of Heterogeneous B2B Services., Proceedings of IEEE CECEEEE 2007. Wache H. et al, 2001. Ontology-Based Integration of Information — A Survey of Existing Approaches. Proceedings of IJCAI-01 Workshop. Seattle, WA, pp. 108–117. Ye C., Cheung S. and Chan W. 2006. Publishing and composition of Atomicity-equivalent Services for B2B Collaboration. Proceedings of ACM ISCE, Shanghai, China, pp. 351–360. W3C, 2005. http://www.w3.org/Submission/WSMX/ W.K. Ng W.K. Yan K.G and Lim E. 2000. Heterogeneous Product Description in Electronic Commerce. ACM SIGecom Exchanges, Vol. 1, No. 1, pp. 7-13. Zhao Y. 2005. Develop the Ontology for Internet Commerce by Reusing Existing Standards. Proceedings of SWAFT’05, Nara, Japan, pp. 51-57. 76 IADIS International Conference e-Commerce 2009 SMART SHOPPING SPACES: CONNECTING MERCHANTS AND CONSUMERS BY INNOVATIVE ONLINE MARKETPLACES Peter Leitner and Thomas Grechenig Research Group for Industrial Software Vienna University of Technology Wiedner Hauptstraße 76, 1040 Vienna, Austria ABSTRACT Accelerated by the big success of leading social web services like Facebook, Flickr, Twitter or YouTube and the request of exciting features through the web community, a new genre of innovative online marketplaces has evolved over the last years. Smart Shopping Spaces provide a unique shopping experience characterized by strong collaboration and social interaction between merchants and consumers. Conventional online shops are gradually updated with innovative shopping features or become fully substituted by new e-shopping concepts, like Fruugo, MyDeco, Threadless or Vente Privée. This paper presents a trendsetting model for Smart Shopping Spaces based on analyzed best practice elements, functionalities and interactions. The designed model combines components of social networking, online shopping and social commerce services. An extended research on 300 individual services out of these three categories was performed to achieve a broad overview. Relevant research results, significant showcases as well as the creation process and an explanation of the model are presented. The framework illustrated in the final part serves as basis for both researchers and developers of future Smart Shopping Spaces. KEYWORDS Smart Shopping Space, Innovative Online Marketplace, Merchant, Consumer, Social Commerce. 1. INTRODUCTION In the age of social web masses of users collaborate, communicate and interact online using innovative services and platforms. Driven by new concepts, technologies and features the web has become more social and interconnected. Social networks like Facebook, social media services like Flickr or YouTube and micro blogging services like Twitter are popular examples of this genre, which strongly changed the web. O`Reilly (2005) summed it up by the buzzword “Web 2.0” and many other wrote about the phenomenon of social web and the wisdom of crowds (Surowiecki, 2004; Howe, 2006). Nowadays every user is an active part of that sphere and can easily become an author, publisher or media distributor (McLure Wasko and Faraj, 2000). Reasons for that phenomenon beside technological innovations and new concepts are the strong increase of internet population worldwide, higher internet bandwidths and the rapid distribution of electronic user devices and gadgets like mobile phones or digital cameras. For example actual data of Facebook show that it currently has about 200 million active users with about 850 million photos and 8 million uploaded videos each month. An average user has 120 friends on the site and about the half of all users log on to Facebook at least once each day. Accelerated by the big success of leading social web services like Facebook, Flickr, Twitter or YouTube and the request of exciting features through the web community a new genre of innovative online marketplaces evolved over the last years. In consequence of social networks’ market power and their mass of potential customers, new B2C and C2C shopping concepts (Schubert and Ginsburg, 2000; Füller et al., 2006) are being developed. Also the request of the web community and consumers for more participation and transparency is a driving factor of these new concepts. Literature (Buskens, 2002, Koivumäki et al., 2002; Kim and Srivastava, 2007) and specific user surveys (Lorenzo et al., 2007; The Nielsen Company, 2007) clearly outline that potential customers of a product attach more importance to recommendations and ratings 77 ISBN: 978-972-8924-89-8 © 2009 IADIS of other users than to classical product descriptions and advertisements. Smart Shopping Spaces are providing such a unique shopping experience characterized by strong collaboration and social interaction between merchants and consumers. Conventional online shops are gradually updated with innovative shopping features or become fully substituted by new e-shopping concepts, like Fruugo, a promising Scandinavian startup which launched its social shopping platform a few weeks ago in the second quarter of 2009. MyDeco, an already successful startup founded 2007 in the UK, is a collaborative interior design network where rooms can be styled in 3D, designs can be shared with other users, and finally they can buy the products and furniture over included home stores. Currently MyDeco has about 770 home stores with more than 3.8 million items. Also Threadless is a pioneer with a Smart Shopping Space concept and changed the t-shirt print business in an impressive way. The basic idea of Threadless is crowdsourcing, as it counters on the increasing heterogeneity of needs by putting the consumers actively into the value added chain. Every user can submit his own print design, the community decides and finally provide product photos through the platform. Another interesting online clothing store concept is Vente Privée, a company founded in France with several spin-offs in other countries. Vente Privée is the pioneer of online shopping clubs, where only exclusive members can buy cheap designer fashion and other products after an invitation of another member. Currently Vente Privée has about 7 million members in Europe and sold 28 million products with revenue of 510 million Euro in 2008. Only in France there are over 50 copycats of that successful business model. These examples exemplify the rise and success of Smart Shopping Spaces, even though we are just on the beginning of this era of new online shopping concepts. As a consequence, the time of monotony in online retail is a thing of the past. More and more merchants and service developers are launching such new shopping models, and smart shopping features will become a must-have for every shop owner in near future (Burt and Sparks, 2003; Machado, 2005). To enable next generation shopping, new social software service models are necessary to meet the demands of innovative merchants and contemporary consumers. This paper presents a trendsetting model for Smart Shopping Spaces based on analyzed best practice interactions, functions and components. The designed model combines elements of social networking, online shopping and social commerce services. An extended research on 300 individual services out of these three categories was done to gain fundamental insights. The following section provides related work, mostly concentrating on academic research and basic literature. Section 3 outlines the methodological approach we defined for our work. In section 4 we present relevant research results and significant showcases of our analyses. Section 5 shows the creation process and an explanation of the Smart Shopping Space model. We finally present a conclusion in section 6 and outline future directions of research in the final section. 2. RELATED WORK Basic literature about Web 2.0 (O'Reilly, 2005; Mikroyannidis, 2007; Chi, 2008) and innovative social web services (Abel et al., 2007; Ankolekar et al., 2008) as well as special publications pointing out sociological and critical aspects (Dron and Bhattacharya, 2007; Riedl, 2008; Yesilada and Harper, 2008) provide an overview on current internet evolutions. Especially Surowiecki (2004) analyzed the wisdom of crowds, an aggregation of information in groups, resulting in decisions that are often better than any decision of a single member of the group. Also crowdsourcing is a neologism, which was defined by Howe (2006) and describes, contrary to outsourcing, the outsourcing to the intelligence and the manpower of voluntary staff on the internet. The phenomenon of social networking and its driving are topic of related literature (Backstrom et al., 2006; Kumar et al., 2006). Particularly Boyd and Ellison (2007) illustrated the global history and impact of social networks. Despite of significant interest in academic circles, as represented in various blog posts (Tedeschi, 2006; Rubel, 2006; Beisel, 2006), little academic research has been done in the field of future Smart Shopping Spaces to date. Work we published recently introduces social commerce (Leitner and Grechenig, 2007) and includes earlier research on social shopping services (Leitner and Grechenig, 2009). To create our Smart Shopping Space model, we discussed different related work in the field of electronic commerce (Treese and Stewart, 2002; Meng et al., 2004; Moreno et al., 2004; Yang and Mamadou, 2006), community design (Sanderson, 1997; O'Day et al. 1998; Pipek, 1999; Kim, 2000; Andrews, 2002; Krieger and Müller, 2003; Breslin et al., 2005) and specific social features and interactive web elements (Yamamoto et al., 2005; Michaelides and Kehoe, 2007; Miyoshi et al. 2007; Szomszor et al., 2008). 78 IADIS International Conference e-Commerce 2009 3. METHODOLOGICAL APPROACH Subsequent to fundamental review of related literature and work we defined a structured methodological approach (Figure 1) to design an innovative model for Smart Shopping Spaces connecting merchants and consumers. Based on a conceptual knowledge of social software services, especially for social commerce, we concretized the main research question to have a strong focus during the different stages of our work. Beside the question on recent components, functionalities and interactions of a modern Smart Shopping Space, we were also interested in future trends and used revenue models in the field of social shopping in general. Figure 1. Structured Methodological Approach 3.1 Selection and Classification To gather insights into relevant components, functionalities and interactions for the Smart Shopping Space model, we categorized related services. Additionally we identified three basic genres of services which are significant for the design of our model. As described in the introduction, a Smart Shopping Space combines the advantages of social networks and online shops under consideration of interactive social commerce features. Thus, we targeted conventional social networking services and online shopping services, as well as social commerce services to conduct our survey. For every of the three service categories we selected 100 services to start extended analyses. The process of selection was performed in an analogous manner for all of the three categories. We collected services from all around the world from well known online resources and web lists like Crunchbase.com, Mashable.com, or Web2List.com. Sorting out services in an early beta stage from our initial list, criteria for the selection were technological maturity, a critical mass of users and a minimum of complexity of the services. Out of the rest we took a random sample of 100 services. So we got a new list of 300 different services which changed partially to an initial list of previous research (Leitner and Grechenig, 2009) and separated them into three different categories to conduct next steps of research. 79 ISBN: 978-972-8924-89-8 © 2009 IADIS 3.2 Evaluation and Summaries To evaluate selected services, a structured spreadsheet (Figure 1) for each service category was created. Based on a specified metric, every service was analyzed, rated and described. Beside some comparable meta data (e.g. year of foundation, location, market orientation, multilingualism etc.), standardized qualitative assessment criteria (e.g. actuality, content, design, functionality, usability) and several feature-related criteria (e.g. profiles, blogs, groups etc.) were defined for each of the three different categories. Qualitative criteria were rated with numbers ranging from 1 to 5, whereas a 5 was the best. Feature-related criteria have been considered with “yes” or “no”. Concrete criteria and results are presented in the analyses section in this paper. After the definition of the metric, services were screened and evaluated. To give a better presentation, we transferred the results of the performed evaluation to a standardized case summary collection, including one page for every analyzed service. Additionally we included a short written description and two screenshots of every observed service on each page. A case book with 300 pages divided in three categories including all services was the result of these steps and built the base for future research and development. 3.3 Analysis and Insights After evaluation of all services and summarization of gathered data in case summaries, we started a detailed analysis (Figure 1) by creating rankings and charts. Main goal was the indication of applicable components and functionalities of each different service category and their concrete percentage of usage. Furthermore we analyzed the interactions between the components and external applications or third party services as well as qualitative criteria. In a final step we documented all results and interpreted rankings and charts to provide an informative basis for the following stages of development. Especially the frequency of usage in percent of distinct components, functionalities and interactions was a significant source for the design process beside other fundamental research on social software service concepts and functionalities, described in the related work section of this paper. 3.4 Entities and Design Before we started with the creation of the smart shopping space model, we grouped all identified components and functionalities into three different parts, similar to the aforementioned service categories. Additionally we defined a centric entity (consumers, products, merchants) for each part of our Smart Shopping Space model. Based on every main entity, we started to build related sub entities, components and functionalities, and finally we connected all elements trough interaction paths. The modular design of our shopping community was an initial condition to assure its scalability and flexibility to integrate it into already existing conventional e-shops or communities. 4. ANALYSES, RESULTS AND SHOWCASES Three analyses of the initially identified service categories, social networking, online shopping and social commerce, were done simultaneously in a similar way. The following part of this paper presents the analysis design, the selected services and results focused on the top ten components and functionalities for each investigated category by their percentage of usage. Additionally one significant showcase for each genre is described to demonstrate relevant components and functions. 4.1 Analysis of Social Networking Services Out of all collected conventional community and social networking services we took a random sample of 100 services (Figure 2) for a standardized evaluation. Beside elements (components, functionalities) and qualitative criteria (actuality, content, design, features, usability) shown in the next section, analyzed criteria were basic data like the year of foundation, market orientation, used revenue models, and multilingualism. 80 IADIS International Conference e-Commerce 2009 Figure 2. Elements and Qualitative Criteria of all Social Networking Services (n=100) Results show that 34% of all social networks have multilingual services, 94% have a global orientation and 24% are closed networks, meaning that a user needs an invitation from another person to join a social networking service. Elements and Qualitative Criteria: The analysis of 100 conventional social networks and communities (Figure 2) outlines that all services (100%) are using customizable user profiles followed by photos (86%) and groups (63%). 56% of all services are using blogs and 52% have integrated forums. 50% of all services have integrated an event calendar and nearly the half video features (48%). About one third of all investigated services are using widgets (31%). Music (17%) and application programming interfaces (10%) to integrate external applications are not very commonly used. In general, most of the analyzed services offer a standard setup with user profiles, photo galleries and forums or groups for communication. Qualitative criteria show high actuality with an average value from 4.54 out of 5 and also values in a high range for content (4.01), design (3.89), features (3.75) and usability (3.94). Shwocase Xing.com: Xing is a business social networking service with more than 7 million registered members worldwide. It was founded in August 2003, launched in November 2003. Xing offers the system as well for closed communities, i.e. private clubs, academic clubs or corporate clubs with own access paths and own interface designs. The platform serves as infrastructure for corporate groups, including Accenture, IBM, McKinsey, and others. Premium groups are available for major global communities, from university and corporate alumni groups to leading business magazines and multinational organizations. Xing is a competitor of the American platform LinkedIn for social networking among businesses. The platform offers personal profiles, groups, discussion forums, event coordination, and other common social community features. Users can build their own profile and make business contacts or connections to other members. There is the possibility to search for other users and to find new contacts or connections. Extended features are available for premium members, who pay a monthly fee. Beside that membership fees Xing generates revenue from onsite advertising. 4.2 Analysis of Online Shopping Services To run an evaluation on conventional online shopping services, we selected 100 services (Figure 3) by a random sample out of a large list of collected sites. The selected services belong to different market segments like fashion, lifestyle, electronics or sports. We focused on components as well as functions and found out that 29% of the selected objects have multilingual services, whereas only 27% have a global market. 21% of all services are certified online shops. 81 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 3. Elements and Qualitative Criteria of all Online Shopping Services (n=100) Elements and Qualitative Criteria: Figure 3 shows the distribution of relevant elements of all 100 analyzed classic online shopping services. The analysis of selected online shops and B2C e-commerce platforms (Figure 3) shows that almost all services (99%) are using photo features, followed by user logins (94%) and onsite search functions (87%). 69% of all services integrated a newsletter feature. User-oriented tools like product viewers and rating tools are implemented within 36% of all services. Blogs are integrated on 20% of all services, 17% have implemented videos or multimedia components. Less than 10% of all investigated shops are using forums (9%) and widgets (7%) for communication and syndication. Qualitative criteria are between 3.12 and 4.49 out of maximum of 5, with high actuality and some feature optimization potential. Showcase Endless.com: Amazon launched Endless.com in response to customers' desires to shop a destination dedicated to shoes and handbags in the year 2007. The Endless e-shop provides an enhanced browsing experience for consumers. Shop visitors may search by category, color, size, brand, or price and alternative views and innovative zoom technology allow a virtually touch of new shoes and handbags. Consumers can navigate easily and quickly to make their choice. For example, if a shopper navigates to black sandals, and chooses a product within that category, similar shoes are displayed above the product he is looking at. Furthermore a customer is only one click away from details of the product which is a great time saver. An exciting tool is the sliding price chooser. As opposed to selecting shoes from a predetermined price range, consumers can actually set their own price points, by just sliding a bar on the product overview. The modern and light design features are realized with innovative technologies and concepts like AJAX. Services like free overnight shipping, free return shipping, price guarantee, and 365-day returns window are offered. Another interesting point is that existing Amazon customers can sign-in using their Amazon.com account. Within the web community there were many positive reactions based on the new shopping experience. Many other new online shops were inspired by the core components, functions and views of Endless.com. 4.3 Analysis of Social Commerce Services Similar to the previous analyses, we took a random sample of 100 social commerce services (Figure 4) for our third evaluation. Beside core elements and qualitative criteria, we investigated the year of foundation, market orientation, used revenue streams, and multilingualism. Results show that only 13% of all social commerce services are multilingual, and 68% have a global market orientation. Elements and Qualitative Criteria: The investigation of 100 innovative social commerce services (Figure 4) outlines that the most used elements are member profiles (96%), photos (64%) and blogs (63%). 82 IADIS International Conference e-Commerce 2009 Figure 4. Elements and Qualitative Criteria of all Social Commerce Services (n=100) Tags for marking products are used in 33% of all analyzed services. Useful collaboration and communication elements like forums (31%) and groups (25%) are integrated as well as widgets (19%), application programming interfaces (14%), videos (10%) and events (8%) within the analyzed services. Qualitative criteria are all in the same range between 3.38 (actuality) and 3.72 (design) out of 5, much more above the ones from conventional online shopping services. Showcase MyDeco.com: is a collaborative interior design network where rooms can be styled in 3D and the designs can be shared with other users. MyDeco was founded in February 2007, in London. Currently there are about 40 employees. The user can directly access to resources of over 770 vendors and more than 3,8 million products. Users can build a 3D view of their flat via drag and drop, to visualize potential purchases and views from different angles in advance. Besides a selection of furniture and accessories, which can be bought as well, there is the possibility to select different colors, wallpapers and floors. To offer an easier shopping experience, there is also the possibility to choose a complete configured adjustment including the whole shopping list and a budget check. Personal designs can be saved and rated from the community members. MyDeco offers some common community features like an own profile, personal data, blogs, groups and the possibility to communicate with other members. MyDeco is not directly selling any of the furniture, but instead works as an intermediary, taking a cut from every sale the site generates for its retail partners. Furthermore the revenue is done by selling advertisement spaces on the website. And MyDeco has an additional micro-affiliate model. Any small interior design business or an individual can upload a room design. If someone takes a design of another member and purchases the items, the designer gets a provision. 5. MODEL FOR SMART SHOPPING SPACES The model for Smart Shopping Spaces (Figure 5) was designed based on our research on components, functions and interactions of all analyzed services, presented in the section before. Also relevant related third party research regarding social networks (e.g. Backstrom et al., 2006; Kumar et al., 2006; Safar and Bin Ghaith, 2006; Srivastava, 2006; Boyd and Ellison, 2007), community design (e.g. Sanderson, 1997; O'Day et al. 1998; Pipek, 1999; Kim, 2000; Andrews, 2002; Krieger and Müller, 2003; Breslin et al., 2005; Skok, 2005), electronic commerce features (e.g. Treese and Stewart, 2002; Meng et al., 2004; Moreno et al., 2004; Yang and Mamadou, 2006), as well as innovative social web elements (e.g. Yamamoto et al., 2005; Mor Naaman et al., 2006; Bao et al., 2007; Michaelides and Kehoe, 2007; Miyoshi et al. 2007; Heymann et al., 83 ISBN: 978-972-8924-89-8 © 2009 IADIS 2008; Szomszor et al., 2008) was considered during the design process. Smart in that context relates to the generation of a new shopping experience for all consumers within a generic framework which can be either Figure 5. Designed Model for Smart Shopping Spaces used to build completely new services or to expand already existing projects, like conventional social networks or online shops. Merchants and service developers should choose their combination of components, functions and interactions with different complexity and granularity based on case-specific requirements. Furthermore the model should be seen as a fundament to generate derivates or as an inspiration for new concept designs. Describing the main business process of a Smart Shopping Space – consumers buy products from merchants –, we defined the following three different main entities (Figure 5) for our model: Consumers: Consumers are the centric entity of the social network part. It is similar to a user in a conventional social network or community. After joining a Smart Shopping Space, the main consumer goals are to interact and collaborate with others in a social way. That means to get advice from trusted individuals, discuss about products, provide related media and probably find new friends online. Beside social interaction, consumers use these platforms to find and finally purchase appropriate products. Products: The products are the second main entity of our framework. In a fully integrated social shopping community it connects the two other main entities by using modern social commerce components and functions. Products have to be clearly categorized and sorted to allow fast access from both directions. The application of tagging, rating and ranking concepts allows user-driven categorization and includes the power of the whole community within one platform. 84 IADIS International Conference e-Commerce 2009 Merchants: Main goals of merchants, the third main entity of the model, within a Smart Shopping Space are to address potential consumers with marketing campaigns and finally sell his products. The merchant, being the central entity of the online shop part, is also responsible for delivering informative content about products and to set them into the right context. To be successful, the vendor has to integrate himself within the whole community like a conventional member. In an optimal way the merchant should collaborate and interact continuously with other consumers to be an active participant. After definition of the three main entities we clustered all identified components and functions of the three analyses done before and positioned relevant ones as sub entities around the three main entities. Besides integrating our own results, we tried to consider important research and aspects presented in the related work section of this paper. The outcome was a simplistic model structure (Figure 5) with three main parts. Around the consumer we built a social networking part, because of its similarity to classic social networks. The online shopping part around the merchant includes main e-shop components and functions. And finally the connecting part between the two other parts is the social commerce part, which is the most innovative part of the model because of an integration of many social web features. Finally we connected all elements with interaction paths to demonstrate their relation. This interaction model clearly shows the complexity of a fully integrated social shopping community. Essential elements of each part are described below. 5.1 Social Networking Part Consumers will join a Smart Shopping Space due to campaigns, the invitation of a friend, a search result of a price comparison engine. After joining the consumer can customize his own profile, run a blog or create his own product wish lists, which are shared with other members on the platform. Users of Smart Shopping Spaces arrange themselves in different groups and find friends or other users with similar interests. All members of a group have the possibility to discuss on products or other related topics. Most of the components and functions are similar to a conventional social network. It is also possible to include premium features for special user groups. 5.2 Online Shopping Part The components and functions of the online shopping part allow merchants to start marketing campaigns to target potential customers, use specific promotion tools and communicate newsletters. All products of a vendor are part of his repository which can be managed and categorized over the backend. Furthermore a merchant has a direct interface for his suppliers to update relevant product information and to provide a direct link between a product and a supplier for consumers. A merchant may have a corporate blog to inform consumers about relevant news. 5.3 Social Commerce Part The social commerce part, connecting the social network and the online shopping part, includes the most innovative components and functions of all parts. Consumers and merchants can use these product-related features to collaborate and interact in a new way. Products can be tagged by members of the shopping community to allow a user generated categorization. Consumers write recommendations or comments on products, rate them and create rankings to show others their preferred products. Social media like interactive photos or videos can be included by consumers or merchants. Question and answers about products or shopping-related topics can be run consecutively between consumers and merchants. The whole collaborative interaction cycle should be seen as a continuous process. After buying a specific product, the user will recommend it within the community, share information with other users, and perhaps they will put it on their wish list. For the whole framework an external connection with the system can be established via several interfaces. Thus, the designed model can be opened for third party applications and as a consequence, communication within the global web sphere is possible. For example, integrated Feeds which are driven by RSS or Atom can be tracked in real time, or interactive product lists can be searched over product search engines and other social web services. Integrated application programming interfaces should provide data interchange with mashups (Leitner and Grechenig, 2008) and syndication with leading social web services. 85 ISBN: 978-972-8924-89-8 © 2009 IADIS 6. CONCLUSION In the era of social web masses of users communicate, collaborate and interact online using innovative services and platforms. Driven by new concepts, technologies and features the web has become more social and interconnected. Forced by the huge success of leading social web services like Facebook, Flickr, Twitter or YouTube and the request of exciting features through the consumer community, a new genre of innovative online marketplaces has evolved over the last years. Such Smart Shopping Spaces provide a unique shopping experience characterized by strong collaboration and social interaction between merchants and consumers. Conventional online shops are gradually updated with innovative shopping features or become fully substituted by new e-shopping concepts, like Fruugo, MyDeco, Threadless or Vente Privée. As outlined in the introduction of this paper, consumers demand recommendations by other users and a direct conversation and interaction with others. Smart Shopping Spaces are combining many of these new demands in one place, allowing consumers to collaborate online, get advice from trusted individuals, find products and then purchase them. Building a successful shopping platform requires a perfect combination of different elements to meet the demands of merchants and consumers. The illustrated Smart Shopping Space model in this paper was designed under consideration of best practice functions, components and interactions of already existing web services. Structured research on social networking, online shopping and social commerce services built a basic framework for the development process of our model. For each of the three categories we took a random sample of 100 services to gather significant information on the usage of distinct components, functions and interactions. The results of each analysis led us to an illuminative ranking of all relevant elements. Results show that the distribution of elements for social networking services is similar to the one of social commerce services, but clearly show some significant differences like the intensified implementation of tags within the second group. Qualitative criteria for actuality, content, design, features and usability show that social networking services have the best average values, followed by social commerce services and conventional online shopping services. There were also significant differences relating to the grade of global orientation and multilingualism between the three different genres of analyzed web services. By grouping all identified elements around three core entities (consumers, products and merchants) and the inclusion of relevant third party research on community design, electronic commerce and social web features, the three-part model was built. Subsequently we connected all entities, sub-entities and components through interaction channels to demonstrate the coherences. The presented model can be used to build innovative shopping concepts and services from two directions with different granularity. One scenario is to upgrade an already existing conventional social network or online shop with smart features. Another scenario is to create completely new shopping services or platforms up from scratch. Our general framework should serve as fundamental basis for researchers, merchants and service developers of trendsetting marketplaces. 7. FUTURE WORK As previously outlined, little academic research has been done on Smart Shopping Spaces and its usage up to date. Future work should cover several important areas relating to such services. One essential research area will be the impact of smart shopping features on consumer’s behavior, especially the influence on revenue of a shop. So there have to be implemented tools and processes on existing e-shops and platforms to measure the impacts. Another interesting area of research will be consumer experience tests on specific features and mass user behavior to get better insights into usability issues of innovative shopping features and concepts. On the technical side there is much potential for the optimization and integration of functionalities, components and tools within the whole web sphere. For example there are currently no standards for the syndication of product data, prices and other shop related data in a bigger context. That means it would be interesting to design possible interfaces and information standards for better communication of different platforms and global distribution of data. Our own future work will be to create different types of Smart Shopping Spaces out of free open source online shop frameworks in combination with our designed model to find the best set of features. In another research project we will cooperate with a leading online merchant to optimize the existing platform by tracking usage data and user behavior. 86 IADIS International Conference e-Commerce 2009 REFERENCES Abel, F., Frank, M., Henze, N., Krause, D., Plappert, D. and Siehndel, P., 2007. GroupMe! - Where Semantic Web Meets Web 2.0. Proceedings of the 6th International Semantic Web Conference, 2nd Asian Semantic Web Conference, ISWC 2007 + ASWC 2007, Busan, Korea, LNCS 4825, Springer, Berlin, pp. 871-878. Andrews, D. C., 2002. Audience-specific online community design. Communications of the ACM, Volume 45, Issue 4, ACM, New York, pp. 64-68. Ankolekar, A., Krötzsch, M., Tran, T. and Vrandečić, D., 2008. The two cultures: Mashing up Web 2.0 and the Semantic Web. Journal of Web Semantics: Science, Services and Agents on the World Wide Web, Volume 6, Issue 1, pp.70-75. Backstrom, L., Huttenlocher, D., Kleinberg, J. and Lan, X., 2006. Group formation in large social networks: Membership, growth, and evolution. Proceedings of 12th International Conference on Knowledge Discovery in Data Mining, ACM Press, New York, pp. 44-54. Bao, S., Xue, G., Wu, X., Yu, Y. and Su, Z., 2007. Optimizing Web Research Using Social Annotations. Proceedings of the 16th International Conference on World Wide Web, Alberta, Canada, ACM Press, New York, pp. 501-510. Beisel, D., 2006. The Emerging Field of Social Commerce and Social Shopping, http://www.genuinevc.com/archives/ 2006/11/the_meaning_of_2.htm Boyd, D. M. and Ellison, N. B., 2007. Social network sites: Definition, history, and scholarship. Journal of ComputerMediated Communication, 13(1), article 11. http://jcmc.indiana.edu/vol13/issue1/boyd.ellison.html Breslin, J., Decker, S. and Harth, A., 2005. An Approach to connect Web Based Communities. Proceedings of IADIS International Conference Web Based Communities 2005, Algarve, Portugal, pp. 272-275. Burt, S. and Sparks, L., 2003. E-commerce and the retail process: a review. Journal of Retailing and Consumer Services, Vol. 10, pp. 275-286. Buskens, V., 2002. Social Networks and Trust, Springer, Berlin, Germany. Chi, E. H., 2008. The Social Web: Research and Opportunities. Computer Journal, Volume 41, Issue 9, Palo Alto: IEEE Computer Society, pp. 88-91. Dron, J. and Bhattacharya, M., 2007. Lost in the Web 2.0 Jungle. Proceedings of the Seventh IEEE International Conference on Advanced Learning Technologies 2007, Niigata, Japan, IEEE Computer Society, pp. 895-896. Füller, J., Bartl, M., Ernst, H. and Mühlbacher H., 2006. Community based innovation: How to integrate members of virtual communities into new product development. Electronic Commerce Research, Volume 6, No. 1, pp. 57-73. Heymann, P., Ramage, D. and Garcia-Molina, H., 2008. Social Tag Prediction. Proceedings of the 31st annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Singapore, ACM Press, New York, pp. 531-538. Howe, J., 2006. The Rise of Crowdsouring, Wired Mag., USA, http://www.wired.com/wired/archive/14.06/crowds.html Kim, A., 2000. Community Building on the Web: Secret Strategies for Successful Online Communities. Peachpit Press, Berkeley, USA. Kim, Y. A. and Srivastava, J., 2007. Impact of Social Influence in E-Commerce Decision Making. Proceedings of the Ninth International Conference on Electronic Commerce, Minneapolis, MN, USA, August 19 - 22, 2007, ACM Press, New York, pp. 293-302. Koivumäki, T., Svento, R., Perttunen, J. and Oinas-Kukkonen, H., 2002. Consumer Choice Behavior and Electronic Shopping Systems: A Theoretical Note. NETNOMICS Journal, Vol. 4, No. 2, pp. 131-144. Krieger, B. L. and Müller, P. S., 2003. Making Internet Communites Work : Reflections on an Unusual Business Model. ACM SIGMIS Database Journal, Volume 34, Issue 2, ACM Press, New York, pp. 50-59. Kumar, R, Novak, J. and Tomkins, A., 2006. Structure and evolution of online social networks. Proceedings of 12th International Conference on Knowledge Discovery in Data Mining, ACM Press, New York, pp. 611-617. Leitner, P. and Grechenig, T., 2007. Next Generation Shopping: Case Study Research on Future E-Commerce Models, Proceedings of the IADIS International Conference E-Commerce 2007, Algarve, Portugal, pp. 312-316. Leitner, P. and Grechenig, T., 2008. Electronic Commerce Mashups: Combining Applications to Useful Services for Shoppers. Proceedings of IADIS International Conference E-Commerce 2008, Amsterdam, Netherlands, pp. 260-264. Leitner, P. and Grechenig, T., 2009. Scalable Social Software Services: Towards a Shopping Community Model Based on Analyses of Established Web Service Components and Functions. Proceedings of the 42nd Annual Hawaii International Conference on System Sciences 2009, Big Island, Hawaii, USA, p. 45, CD-ROM 10 pages. Lorenzo, C., Constantinides, E., Geurts, P. and Gómez, M. A., 2007. Impact of Web Experience on E-Consumer Responses. Proceedings of the 8th International Conference EC-Web 2007, LNCS 4655, Regensburg, Germany, Springer, Berlin, pp. 191-200. Machado, A., 2005. Drivers of Shopping Online: A Literature Review. Proceedings of IADIS International Conference e Commerce 2005, Porto, Portugal, pp. 236-242. 87 ISBN: 978-972-8924-89-8 © 2009 IADIS McLure Wasko, M. and Faraj, S., 2000. “It is what one does”: Why people participate and help others in electronic communities in practice. Journal of Strategic Information Systems, No. 9, pp. 155-173. Meng, B., Zhang, H. and Xiong, Q., 2004. Design and Implementation of a Business to Customer Electronic Commerce System applying Web Technologies. Proceedings of IADIS International Conference e-Commerce 2004, Lisbon, Portugal, pp. 527-532. Michaelides, R. and Kehoe, D., 2007. Internet Communities and Open Innovation: an Information System Design Methodology. Proceedings of the 6th IEEE/ACIS International Conference on Computer and Information Science 2007, Melbourne, Australia., IEEE Computer Society, pp. 769-775. Mikroyannidis, A, 2007. Toward a Social Semantic Web. Computer Journal. Volume 40, Issue 11, IEEE Computer Society, Leeds, pp. 113-115. Miyoshi, Y. Kanenishi, K., Matsuura, K., Ide, R. and Yano, Y, 2007. A Proposal to extend Folksonomy based on Tags and Links. Proceedings of the IADIS International Conference Web Based Communities 2007, Salamanca, Spain, pp. 389-392. Mor Naaman, M., Boyd, D. and Davis, M., 2006. HT06, Tagging Paper, Taxonomy, Flickr, Academic Article, ToRead. Proceedings of Hypertext 2006, ACM Press, New York, pp. 31-40. Moreno Chaustre, J., Naranjo Cuervo, R., Sierra Martinez, L. and Mendoza Becerra, M., 2004. Building a Virtual ECommerce Community. Proceedings of IADIS International Conference e-Commerce 2004, Lisbon, Portugal, pp. 495-500. O'Day, V. L., Bobrow, D. G. and Shirley, M., 1998. Network Community Design: A Social-Technical Design Circle. Computer Supported Cooperative Work (CSCW) Journal, Volume 7, Numbers 3-4, Springer, pp. 315-337. O'Reilly, T., 2005. What Is Web 2.0. Design Patterns and Business Models for the Next Generation of Software. http://www.oreillynet.com/pub/a/oreilly/tim/news/2005/09/30/what-is-web-20.html Pipek, V., 1999. Enabling for mass communication in community design. ACM SIGGROUP Bulletin, Volume 20, Issue 2, ACM Press, New York, pp. 13-15. Riedl, J., 2008. Altruism, Selfishness, and Destructiveness on the Social Web. Proceedings of the 5th International Conference AH 2008, LNCS 5149, Hannover, Germany, Springer, Berlin, pp. 9-11. Rubel, S., 2006. Trends to Watch Part II: Social Commerce, http://www.micropersuasion.com/2005/12/2006_trends _to_.html/ Safar, M. and Bin Ghaith, H., 2006. Friends Network. Proceedings of IADIS International Conference WWW/Internet 2006, Murcia, Spain, pp. 143-146. Sanderson, D., 1997. Virtual communities, design metaphors, and systems to support community groups. ACM SIGGROUP Bulletin, Volume 18, Issue 1, ACM Press, New York, pp. 44-46. Schubert, P. and Ginsburg, M., 2000. Virtual Communities of Transaction: The Role of Personalization in Electronic Commerce. Electronic Markets Journal, Volume 10, Number 1, pp. 45-55. Skog, D., 2005. Social interaction in virtual communities: The significance of technology. International Journal of Web Based Communities, 1 (4), pp. 464-474. Srivastava, K., 2006. Communication-Based Social Networks. Proceedings of IADIS International Conference WWW/Internet 2006, Murcia, Spain, pp. 273-280. Surowiecki, J., 2004. The Wisdom of Crowds. Doubleday Publishing, New York, USA. Szomszor, M., Cantador, I. and Alani, H., 2008. Correlating User Profiles from Multiple Folksonomies. Proceedings of the Nineteenth ACM Conference on Hypertext and Hypermedia, Pittsburgh, USA, ACM Press, New York, pp. 33-42. Tedeschi, B., 2006. Like Shopping? Social Networking? Try Social Shopping. New York Times E-Commerce Report, published on September 11, 2006, http://www.nytimes.com/2006/09/11/technology/11ecom.html The Nielsen Company, 2007. Nielsen Online Global Consumer Study April 2007, http://www.nielsenmedia.com/nc/ portal/site/Public/menuitem.55dc65b4a7d5adff3f65936147a062a0/?vgnextoid=e7cb07dd40b55110VgnVCM100000a c0a260aRCRD Treese, W. and Stewart, L., 2002. Designing Systems for Internet Commerce. 2nd ed. Addison-Wesley Professional, Boston, USA. Yamamoto, H., Ishida, K. and Ohta, T., 2005. Temptation and Contribution in C2C Transactions: Implications for Designing Reputation Management Systems. Trusting Agents for Trusting Electronic Societies - Theory and Applications in HCI and E-Commerce, LNCS 3577, Springer, Berlin, pp. 218-234. Yang, K. and Mamadou, T., 2006. A Conceptual Framework for Semantic Web-Based eCommerce. Proceedings of IADIS International Conference e-Commerce 2006, Barcelona, Spain, pp. 287-291. Yesilada, Y., and Harper, S., 2008. Web 2.0 and the Semantic Web: Hindrance or Opportunity?. SIGACCESS Accessibility and Computing Journal, Issue 90, ACM Press, New York, pp. 19-31. 88 IADIS International Conference e-Commerce 2009 FROM A RETAIL-BASED E-MICROPAYMENT PROGRAM TO A MULTI-PURPOSE PROGRAM: WHICH CRITICAL FACTORS ARE NEEDED FOR SUCH A TRANSFORMATION? Wee Kheng-Tan Assistant Professor, Department of Information & Electronic Commerce, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County, Taiwan Yu-Jie Tan Department of Information & Electronic Commerce, Kainan University, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County, Taiwan ABSTRACT Micropayment is common in retail outlets and public transport system. Numerous electronic micropayment programs have been introduced to reduce the need for users to carry cash. However, many such programs are not successful. Hence there is a need to understand the critical factors which contribute to the success of e-micropayment programs. This study used i-cash; an electronic wallet introduced and used in 7-Eleven, the largest convenience store chain in Taiwan, for such an investigation. Furthermore, i-cash was used for an additional reason. With the recent regulatory relaxation; it is now possible for i-cash to be transformed from a single purpose, retail-based e-micropayment to a multi-purpose one. Through the Decision Making Trial and Evaluation Laboratory (DEMATEL) method, this study shows that network effect plays a very important role in the success of e-micropayment system and is also instrumental in the transformation of a singlepurpose e-micropayment program to a multi-purpose one. The factors influencing the development of e-micropayment programs are highly inter-related and in fact, re-enforce one another, leading to positive feedback and bandwagon effect. It is found that factors such as reliability, more value-add locations, acceptance by merchants and users not only increase the popularity of the system, they also affects and re-enforces other factors through a positive feedback loop. KEYWORDS i-cash; network effect; DEMATEL. 1. INTRODUCTION Micropayment transaction is common in retail outlets and public transport system. The Bank for International Settlements (2001) defines micropayment as a small payment which is uneconomical to process through traditional payment such as credit card as the latter processing charge could exceed the value of the micropayment itself. In order to make such payment, consumers need to use coins and cash, hence causing inconvenience to them. Anticipating business opportunity, operators have introduced a number of smart-card based e-micropayment programs around the world. However, with the exception of a few, such as the Octopus Card of Hong Kong, many such programs are not successful. Hence, from the viewpoint of academic researchers and practitioners, it is useful to determine the factors leading to the successful adoption and diffusion of such programs. i-cash, a popular contactless smart-card electronic wallet system for payment at the 7-Eleven convenience store, is the subject of investigation in this paper. i-cash was launched in 2004 by the convenience store operator to lock-in customer’s loyalty and to increase its sales volume since based on Japan's experience with electronic card, cardholders would spend 10 percent to 15 percent more each time when using the card (Taipei Times, 2007). Users of i-cash are given discount and bonus points when they use i-cash for their purchase. 89 ISBN: 978-972-8924-89-8 © 2009 IADIS Operator of convenience store entering the e-micropayment business reflects the reality of the retail structure of Taiwan. Taiwan has the highest density of convenience stores in the world (CENS.com, 2007a). The island, with an area of 36,000 sq km, has 9,100 convenience stores in 2009 for a population of 23 million (CENS.com, 2009a). 7-Eleven chain dominates with 4,900 stores (53.8%) and its operator, Uni-President Group, is a major conglomerate in Taiwan and operates pharmacy chain store, bakery shops, Starbuck etc. With most of the transactions in convenience stores being micropayment, it is not surprising that this major convenience store chain joins the e-micropayment business. With close to five-million i-cash cards being in use, it is Taiwan second-largest e-wallet product after EasyCard (Taipei Times, 2005; CENS.com, 2007b). EasyCard, a contactless smart-card ticketing system for metro, buses and car parks in Taipei City and Taipei County of Taiwan is widely used by the metro and public bus commuters in Taipei City and its neighborhood. More than 15 million EasyCard have been issued (CENS.com, 2009b) and card users enjoy 20 per cent discount on each metro trip. Until recently, i-cash is a single-purpose e-micropayment program. It can only be used for retail-based payment and extension of i-cash to include other non-retail micropayment transactions e.g. payments for public transport, is restricted by Taiwan’s Banking Act. According to the Act, only banks are allowed to issue cash storage cards with multiple functionalities (Taipei Times, 2006). Furthermore, given that it is introduced by the operator of 7-Eleven, i-cash is predominantly used in 7-Eleven and some of the related retail outlets owned or run by Uni-President. In January 2009, this regulatory restriction was relaxed and approved nonfinancial institutions could now issue stored-value cards with multi-functionalities (Taipei Times, 2009). This relaxation has now opened a window of opportunity for i-cash to broaden its scope of applications to include other non-retail based micropayment transactions, i.e. transforming into a multi-purpose emicropayment program. However, it needs the blessing of consumers to be successful. Given the success of i-cash, it is therefore a suitable candidate to study the critical factors that contribute to the successful development of e-micropayment program. In addition, with i-cash now on the verge of transforming into a multi-purpose program, it may be insightful to examine what are some of the factors which can aid and speed up such a transformation. These factors could be inter-related and influenced one another, hence this research used the Decision Making Trial and Evaluation Laboratory (DEMATEL) method to study which factors are more important than the others and also to examine their inter-relationship. Furthermore, we believe that even though we used i-cash as the subject of investigation, the results could be generalized to other e-micropayment programs. 2. LITERATURE REVIEW Smart Card Alliance named speed, convenience, able to track spending, security and benefits to consumers as drivers of electronic payment system (Vanderhoof, 2007). Madhoushi and Mohebi (2004) listed security, acceptability, convenience, cost, anonymity, control, traceability and control of encryption methods as requirements of electronic payment. Researchers e.g. Linck, Pousttchi and Wiedemann (2006) were also interested with security issues such as encryption and double spending. Hence, whether the electronic wallet is reliable is of concern to everybody. Coase (1937) initiated the idea of transaction cost which covered all the costs arising from conducting a transaction (Williamson, 1985). From the viewpoint of transaction cost, Papaefstathiou and Manifavas (2004) explained that since micropayment systems were used to purchase inexpensive items, it was vital to keep the cost of individual transactions low. Time spent was also a component of transaction cost. Discount could lower the overall cost of acquiring the service or good related to the payment. Convenience is definitely important. Slawsky and Zafar (2005) mentioned that the primary causes for failure of Mondex and other similar schemes were linked to its inability to offer the same level of convenience that payment cards offered for mid-value transactions. Convenience could come in various forms such as placing add-value kiosks in more locations for convenient adding of credit to the e-wallet; allow cash, ATM transfer etc. to add credit to the e-wallet, easy to carry the e-wallet around not just in the form of a card but also as part of accessory such as watch and mobile phone. Technology Acceptance Model (TAM), including the extended forms (Davis, 1989; Davis, Bagozzi and Warshaw, 1989; Venkatesh and Davis, 2000) is widely used to explore why users adopt various systems and applications. One of the major measures of TAM is ease of use which is defined as "the degree to which a 90 IADIS International Conference e-Commerce 2009 person believes that using a particular system would be free from effort" (Davis, 1989). Success of e-micropayment program depends to a large extent on the presence of network effect (also called network externality). Network externality is defined as an increase in value of a product as the number of users of that product increases (Katz and Shapiro, 1985). Associated with this concept is the idea of networked goods. Networked goods have characteristics such as complementarities, switching cost, lock-in, and economies of scale (Shy, 2001). Electronic payment program is an example of a networked good (Baddeley, 2004) with its market being a two-sided market, both consumers and merchants subjecting to network effects and facing the chicken and egg dilemma (See-To, Jaisingh and Tam, 2007). One key characteristic which can contribute to network effect is the presence of a captive audience that drives critical mass (Chakravorti, 2004). A program which is frequently used by the consumers will bring about high transaction volume leading to economics of scale (Rochet and Tirole, 2003). Public transport system is one good choice for e-micropayment program as it has the required captive audience. If these consumers are concentrated in a small geographical region, chance of success will be even higher (Van Hove, 2004). Launched in 1997, Hong Kong’s Octopus card has many of the above characteristics. The card has outperformed other e-micropayment programs backed by international financial organizations (Westland et al., 1997). Octopus card has the support of key transport companies (Poon and Chau, 2001). With more than 70 percent of Hong Kong residents using public transport everyday and concentrating in a small geographical region, it has the captive market to achieve critical mass. Using contactless smart card, it is reliable and completes transactions faster than cash as well as offers an automatic reload feature. It also offers convenience and benefits to its users. Over 400 product and service providers (e.g. F&B and recreational outlets, supermarket, and convenience stores), accept Octopus card. 3. RESEARCH METHOD - DEMATEL Developed by the Battelle Memorial Institute of Geneva, Decision Making Trial and Evaluation Laboratory (DEMATEL) is used to analyze the management problems of complex and inter-related relationship. It can also convert the relationship between the causes and effects of factors into an intelligible structural model of the system (Hung, Chou and Tzeng, 2006; Tzeng, Chiang and Li., 2007). DEMATEL method can be summarized in the following steps: Step 1: Find the average matrix. Suppose there are n factors to consider and there are H respondents in this study. Each respondent is asked to state the degree he or she believes a factor i affects factor j through a score ranging from 0 to 4, with 0 indicating ‘no influence’ and 4 indicating ‘very high influence’. We will get H answer matrix where each answer matrix is an n x n matrix Xk with 1 ≤ k ≤ H. The initial direct relation matrix A is obtained as below: a ij = 1 H H ∑x k =1 k ij (1) Step 2: Calculate the normalized initial direct-relation matrix D. The matrix D is obtained by normalizing initial direct relation matrix A: D= A s (2) where ⎛ ⎜ 1≤i ≤ n ⎝ s = max ⎜ max n ⎞ a , max a ij ⎟⎟ ∑ ∑ ij 1≤ j ≤ n j =1 i =1 ⎠ n (3) Step 3: Calculate the total relation matrix T. A continuous decrease of the indrect effects of problem along the powers of matrix D, e.g. D2, D3, ... , D∞ guarantees convergent solutions to the matrix inversion similiar to an absorbing Markov chain matrix. The total relation matrix T is a n x n matrix and is defined as follows: T = D + D2 + … + Dm = D(I －D)-1 (4) 91 ISBN: 978-972-8924-89-8 © 2009 IADIS as m → ∞ and I is a n x n identity matrix. The sum of rows r and the sum of columns c of the total relation matrix T is obtained as follow: ⎛ n ⎞ ⎟ ⎜ ∑ t ij ⎟ j = 1 ⎝ ⎠ nx1 r = [ri]nx1 = ⎜ (5) ' c= ' [cj] 1xn ⎛ n ⎞ = ⎜ ∑ t ij ⎟ ⎝ i =1 ⎠1xn (6) The sum ri + ci gives an index representing the total effects both given and received by factor i. The difference ri－ci shows the net effect factor i contributes to the problem. If the difference ri－ci is positive, factor i is a net causer, and when ri－ci is negative, factor i is a net receiver. Step 4: Set the threshold value and draw the influence map. 4. SURVEY AND ANALYSIS In order to find out the suitable factors for analysis, this research conducted literature review, interviewed experts and the users of i-cash. Eleven factors were identified and chosen for DEMATEL analysis. They are: - F1: low transaction cost; - F2: discount; - F3: ease of use; - F4: appear in forms other than card format; - F5: many value-add locations and methods; - F6: reliable; - F7: accepted by many transport-based companies; - F8: accepted by many retail outlets; - F9: accepted by many well-known, large scale corporations and organizations; - F10: usable island-wide; and lastly - F11: broad user base. Survey method was used to solicit the opinion of users of i-cash in two areas. Firstly, they were asked to write down the importance of each factor and also for each factor, the scope for improvement through a score ranging from 1 to 5 with 1 indicating ‘little’ and 5 indicating ‘very’. Secondly, each respondent was asked to state the degree he or she believed an factor i affects factor j through a score ranging from 0 to 4 with 0 indicating ‘no influence’ and 4 indicating ‘very high influence’. The survey lasted for 1.5 month starting from January 15, 2009. A total of 63 valid returns were received, 37 (58.7%) of them were male and 44 (69.8%) were in the main age group of 20-30 year old. In addition, 45 (71.4%) of them received college and university education. 4.1 Importance and Scope for Improvement The average importance score and the average score of scope for improvement of the 11 factors were 4.20 and 3.58 respectively (Figure. 1). Five factors: discount, many value-add locations and methods, accepted by many retail outlets, usable island-wide and lastly broad user base had importance and scope for improvement which were above or equal to their respective average score. Two factors: accepted by many transport-based companies and lastly, accepted by many well-known, large scale corporations and organizations had importance which was below average in importance but with scope for improvement greater than its average score for scope for improvement. Appear in forms other than card format had the lowest importance score and ease of use had the lowest score of scope for improvement. 92 IADIS International Conference e-Commerce 2009 Figure 1. Score of Scope for Improvement Versus Importance Score 4.2 Relationship between Factors The initial direct relation matrix A was first obtained and normalized to obtain the normalized initial directrelation matrix D. Total relation matrix T was obtained using (4). The sum of rows r and sum of columns c of matrix T were then obtained, afterwhich the sum ri + ci and the difference ri－ci were calculated using (5) and (6) with result as in Table 1. Table 1. Sum of Rows (r), Sum of Columns (c), Sum (r + c) and Difference (r－c) of Matrix T Factor F1: Low transaction cost F2: Discount F3: Ease of use F4: Appear in forms other than card format F5: Many value-add locations and methods F6: Reliable F7: Accepted by many transport-based companies F8: Accepted by many retail outlets F9: Accepted by many well-known, large scale corporations & organizations F10: Usable island-wide F11: Broad user base r 5.17 4.83 5.31 4.82 5.53 5.21 5.55 5.47 5.46 5.96 5.90 c 4.94 4.90 5.44 4.39 5.67 5.50 5.38 5.28 5.31 6.11 6.31 r+ c 10.12 9.73 10.75 9.21 11.20 10.71 10.93 10.75 10.77 12.07 12.20 r－c 0.23 -0.07 -0.13 0.43 -0.14 -0.29 0.17 0.20 0.15 -0.14 -0.41 It could be observed that the three factors with the highest ri+ci value: broad user base, usable island-wide and lastly, many value-add locations and methods were all intimately related to the concept of network effect and they also affected the causation relationship between the different factors. These 3 factors were also net receivers with ri－ci being negative value. The top 3 net causer factors were appear in forms other than card format, low transaction cost and accepted by many retail outlets. The top 4 net receiver were broad user base, reliable, usable island-wide and lastly, many value-add locations and methods. Using 0.49, the average value of the elements of matrix T, as the threshold value, the influence map was shown in Figure. 2. 93 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 2. The Influence Map 4.3 Discussion Viewing the above in totality and using i-cash as an illustration, e-micropayment system as a networked good is evident. Furthermore, the opinion of the users at this juncture of i-cash’s transformation into a multipurpose card also provides many useful insights pertaining to the impact of some of these factors to the transformation and their contribution to network effects. When the influence map is examined in greater detail, we observe that many of the eleven factors examined are inter-related and they re-enforce one another. As an example, getting more retail outlets to accept i-cash is not an end itself, it also acts as an impetus and pushes more transport-related companies as well as large-scale corporations and organizations to accept i-cash, which in turn their acceptance encourages even more retail outlets and others to accept i-cash and so on. More users will be drawn in, and over time, positive network effects can create bandwagon effect as the network becomes more valuable and causing even more users to use i-cash, leading to a positive feedback loop. Furthermore, many of these factors, either indirectly or directly, point ultimately towards two factors, usable island-wide and broad user base, which in turn point back to many of these factors. This observation clearly shows that these factors re-enforce one another, as more users and merchants fall within the ambit of i-cash, it will make the system stronger over time. The five factors which the users feel are very important and need lot of improvement: discount, many value-add locations and methods, accepted by many retail outlets, usable island-wide and lastly broad user base are also intimately related to network effects. The starting point and the initial purpose of the e-micropayment system will affect the perception of users for some time to come. It also affects its transformation into a multi-purpose program. In this case, i-cash starts off as a retail-related e-micropayment card. Even though there is a recent regulatory liberalization, users still hold on to their perception that i-cash is meant for retail-related usage. Hence, the outcome where users are of the opinion that acceptance by many retail outlets is more important than being accepted by many transport-based companies arises. In addition, acceptance by retail outlets is also one of the top 3 causers (with the largest positive ri－ci value). However, acceptance by many transport-based companies, well-known, large scale corporations and organizations having importance which are below average in importance but with scope for improvement greater than its average score for scope for improvement serves as an important and useful reminder for i-cash operator i.e. for i-cash to undergo a successful transformation from a retail-based e-micropayment scheme to a multi-purpose scheme (like the Octopus Card of Hong Kong), the operators must also get non-retail merchants and organizations to accept and endorse i-cash. 94 IADIS International Conference e-Commerce 2009 As expected, starting as a retail-related e-micropayment card, offering discount and bonus point when using i-cash are important if not the only deciding critical factor of success. Enjoying discount and bonus point when using i-cash sweeten the deals for users. However, the discount offered is often not given by the card issuers but by the merchants themselves. Hence, the merchants must see good value in this e-wallet whereby i-cash can bring in more customers and deals which otherwise may not be so forthcoming if the merchants don’t accept i-cash. A big pool of i-cash users will thus be important and an incentive for merchants to be aggressive in offering discount. One important point from this study and the influence map is clear, i.e. discount will definitely contribute to the growth of user base and participating merchants. Another factor which can contribute to the expansion of e-micropayment program is proof of being reliable and is a well-tested system. Currently, i-cash is only used in 7-Eleven convenience stores and some of the Uni-President’s related enterprises. It is not yet used in a scale which can closely match its closest rival, EasyCard. Hence, from the analysis and influence map, it can be observed that i-cash needs to spend some effort to prove it is very reliable. Besides expanding to more retail outlets and transport-related companies, one important move is to get big and well-known companies and organizations, e.g. government departments, to accept and endorse i-cash. Such acceptance will be read by the public that i-cash is very reliable, hence allowing network effects to play the role it is supposed to do, leading to acceptance by more organizations, being used island-wide and broadening its user base. What are the lessons for the operator of i-cash? Users are of the opinion that the operator need to increase the number of value-add locations and methods of value-add, offer discount, and attract more retail, transport-related, and large corporations to accept i-cash. As indicated by the influence map, these factors reenforce one another, hence we may say it is a chicken and egg situation. The operator will have no choice but to adopt a multi-pronged approach to tackle this issue so that it will achieve the ultimate objectives: broad user base and island-wide usage. Having an existing captive market and good application with captive audience property do help i-cash in its early stage of development. However, it needs to do much more to have a successful transformation and earn the reputation that it is a very reliable system applicable to a wide range of usage. A way to do so will be to attract large and well-known corporations and government organizations to accept and endorse i-cash. It must also work out a deal with participating merchants so that users will see financial benefits e.g. discount and bonus points when using i-cash to make payment and lower the transaction cost of users as much as possible. Last but not least, the operator will have to launch a marketing campaign to alter the perception of users and merchants that i-cash is only a retail-related emicropayment system if it wants to expand into other non retail-related functions quickly. 5. CONCLUSION Through the DEMATEL method, this study shows that the factors influencing the development of emicropayment programs are highly inter-related and in fact, re-enforce one another, leading to network effect. Through using i-cash as an illustration, an electronic wallet introduced and used in 7-Eleven, the largest convenience store chain in Taiwan, we observe that factors such as reliability, more value-add locations, acceptance by merchants and users not only increases the popularity of the system, it also affects and reenforce other factors through a positive feedback loop. REFERENCES Baddeley M., 2004. Using e-Cash in the New Economy: An Economic Analysis of Micropayment Systems. In Journal of Electronic Commerce Research, Vol. 5, No. 4, pp. 239-253. Bank for International Settlements, 2001. Electronic Finance: A New Perspective and Challenges. BIS Papers No 7. CENS.com, 2007a. Taiwan’s Leading Convenience Chains Slow Down Expansion. Viewed 30 April, 2007, <http://cens.com/cens/html/en/news/news_inner_18976.html>. CENS.com, 2007b. E-wallet Catching on at Domestic Convenience Stores. Viewed 2 September, 2007, <http://cens.com/cens/html/en/news/news_inner_20356.html>. CENS.com, 2009a. Growth Slows for Taiwan Convenience Stores. Viewed 6 January, 2009, <http://www.cens.com/cens/html/en/news/news_inner_25813.html>. 95 ISBN: 978-972-8924-89-8 © 2009 IADIS CENS.com, 2009b. New Law Allows Issuance of E-Wallets by Non-Financial Institutions. Viewed 14 January, 2009, <http://www.cens.com/cens/html/en/news/news_inner_26031.html>. Chakravorti S., 2004. Why has Stored-value Not Caught On? In Journal of Financial Transformation, Vol. 12 (December), pp. 39-48. Coase R., 1937. The Nature of the Firm. In Economica, Vol. 4, No. 16, pp. 386-405. Davis F. D., 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. In MIS Quarterly, Vol. 13, No. 3, pp. 319-340. Davis F. D., Bagozzi R. P. and Warshaw P. R., 1989. User Acceptance of Computer Technology: A Comparison of Two Theoretical Models. In Management Science, Vol. 35, No. 8, pp. 982-1003. Hung Y. H., Chou S. C. T. and Tzeng G. H., 2006. Using a Fuzzy Group Decision Approach-knowledge Management Adoption. APRU DLI 2006 Conference. University of Tokyo, Japan, 8–10 November. Katz M. and Shapiro C., 1985. Network Externalities, Competition and Compatibility. In American Economic Review, Vol. 75, No. 3, pp. 424-440. Linck K., Pousttchi K. and Wiedemann D. G., 2006. Security Issues in Mobile Payment from the Customer Viewpoint. Proceedings of the 14th European Conference on Information Systems (ECIS). Göteborg, Schweden, pp. 1-11. Madhoushi M. and Mohebi E., 2004. Multi-analyses Electronic Payment System in Human Perspectives. In Internet Society: Culture, Psychology and Gender, K. Morgan, C. A. Brebbia, J. Sanchez and A. Voiskounsky, Eds. WIT Press, Southampton. Papaefstathiou I. and Manifavas C., 2004. Evaluation of Micropayment Transaction Costs. In Journal of Electronic Commerce Research, Vol. 5, No. 2, pp. 99-114. Poon S. and Chau P., 2001. Octopus: The Growing e-payment System in Hong Kong. In Electronic Markets, Vol. 11, No. 2, pp. 97–106. Rochet J-C. and Tirole J., 2003. An Economic Analysis of the Determination of Interchange Fees in Payment Card Systems. In Review of Network Economics, Vol. 2, No. 2, pp. 69–79. See-To E., Jaisingh J. and Tam K. Y., 2007. Analysis of Electronic Micro-payment Market. In Journal of Electronic Commerce Research, Vol. 8, No. 1, pp. 63-83. Shy O., 2001. The Economics of Network Industries. Cambridge University Press, Cambridge, UK. Slawsky J. and Zafar S., 2005. Developing and Managing a Successful Payment Cards Business. Gower, Hampshire. Taipei Times, 2005. ‘President Chain to Expand icash Cards’, 8 Aug. Taipei Times, 2006. ‘Announcing 3-in-1 EasyCard’, 18 July. Taipei Times, 2007. ‘FamilyMart Puts Faith in e-wallet’, 5 Mar. Taipei Times, 2009. ‘Expanded Use of Electronic Cards Gets Green Light’, 14 January. Tzeng G. H., Chiang C. H. and Li C. W., 2007. Evaluating Intertwined Effects in e-learning Programs: A Novel Hybrid MCDM Model Based on Factor Analysis and DEMATEL. In Expert Systems with Applications, Vol. 32, No. 4, pp. 1028–1044. Van Hove L., 2004. Electronic Purses in Euroland: Why Do Penetration and Usage Rates Differ? In SUERF Studies, B. Morten, Ed, Société Universitaire Européenne de Recherches Financières, Vienna, Austria. Vanderhoof R., 2007. Contactless Payments: Delivering Merchant and Consumer Benefits. EPRG Contactless Forum, Federal Reserve Bank of Boston. Venkatesh V. and Davis F. D., 2000. A Theoretical Extension of the Technology Acceptance Model: Four Longitudinal Field Studies. In Management Science, Vol. 46, No. 2, pp. 186–204. Westland C., Kwok M., Shu T., Kwok T. and Ho H., 1997. Electronic Cash in Hong Kong. In Electronic Markets, Vol. 7, No. 2, pp. 3-6. Williamson O. E., 1985. The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. Free Press, New York. 96 IADIS International Conference e-Commerce 2009 SITE PERSONALIZATION PROCESS BASED ON NAVIGATIONAL BEHAVIOR AND FUZZY ONTOLOGY Juliano Z. Blanco, Antonio Francisco do Prado Computer Science Department – Federal University of São Carlos Via Washington Luis, km 235, 13565-905- São Carlos ABSTRACT This article presents a process that uses the implicit information from navigational behavior and fuzzy ontology for sites personalization according to the preferences of the users. These preferences are organized into profiles and provide the basis of the personalization process, aiming the improvement of the services for the users. In this process not only the history of navigation and preferences detected or expressed by the user are considered for personalization but also the types of devices and networks used for communication, and even the content presented. The process stands out for directing the architecture of the site to provide personalized services. KEYWORDS Personalization Site, User Navigational Behavior, Fuzzy Ontology, Profiles, Content Adaptation. 1. INTRODUCTION The increase and diversification of content available on many sites and the strategies to deliver this content more adequately contribute to the emergence of new personalization processes. These procedures aim to improve the interaction of the user with the web content, favoring the access to the information of his interest. Procedures of personalization provide to the site competitive advantage in relation with its concurrents, creating new business opportunities through the recommendations of its content. In a market where the search for excellence in service is increasingly growing, the obtainment of important information about your customers, such as area of interest and how they do business, can be a key factor for the success of the site. Today, the use of cell phones, personal digital assistant, smart phones and other devices has increased significantly and must be considered by the companies that offer their products and services on the sites. Motivated by these ideas, this article presents a process to improve service for sites users, building their architectures driven by content personalization. This process obtains the site architecture with a focus on personalization of its content as the users’ preferences, their access and navigation devices, features of the networks, and other information such as regional variations. For site architecture to be driven by personalization, the process of building and updating the architecture must consider the continuous changes on the preferences of their users, the devices, the access networks and the content of the site. 2. FUZZY ONTOLOGY AND PROFILES This section provides an overview of the fuzzy ontology, used for knowledge elicitation and site domain representation. It is also presented the profiles used to contextualize the personalization and the adaptation of content, and applied in the process of personalization exposed. 97 ISBN: 978-972-8924-89-8 © 2009 IADIS 2.1 Fuzzy Ontology According to Gruber (1993) [Gruber, 1993], an ontology is a specification of a conceptualization. Ontology refers to an explicit description of concepts and relations of an domain application. There are several languages to model ontology in the Semantic Web [Berners-Lee et al., 2001]. In this work it is used OWLDL because it has the assurance that its inferences are computable. However, these ontologies only capture accurate or complete information. In personalization of sites there are relationships between their users and contents that are not accurate, requiring mechanisms that allow the representation of the degree in which these relationships occur (pertinence degree μ, between 0 and 1) for an appropriate recommendation of these contents. In the SADP process it is used the Fuzzy Meta-Ontology for representation of imprecise information available on the site and an Inference Module on Fuzzy Ontology [Yaguinuma et al., 2007], called in this work as IMFO. The IMFO retrieves from the fuzzy ontology information about the pertinence degree of the relationships between the contents, enabling the extent of the recommendations to the users, according to their preferences. 2.2 Profiles In personalized sites it shall be considered the differences between users, devices and access networks, to enable the appropriately customization of contents. These differences are obtained analyzing the information collected from the user access on the site, and are represented in profiles of: users; devices; access networks; and content. The User Profile is built based on fuzzy ontology models of the site and on the information collected from users' navigation on the site. In SADP process, the main information collected to develop the profile of the user are: date and time of access, user IP address, content accessed, total access time, number of returnings to the same content and clicks made on the pages (clickstream). The Device Profile defines the hardware and software features of the device used for access. It is created from the identification of the browser used to access the sites. Thus, it is possible to collect the following information from the device profile: brand, model, features of the device screen, memory capacity, and formats of contents that can be displayed on this device. For the Network Profile, the calculating time of sending a package to a user’s machine is used to identify the network latency and the information of the Uniform Resource Locator (URL) to obtain the protocol used. The Content Profile contains the features of the content request, expressed in the URL, as the format of the content in WML, XML, HTML, among others. 3. SADP PROCESS COMPUTATIONAL SUPPORT Two frameworks and three tools automatize a large part of the SADP process. The UBICK framework [Santana et al., 2007] supports the activities of the content adaptation directed to PCs to be displayed on mobile devices like cell phones, PDAs and others. This framework intercepts the requests from users and makes the adjustments based on the device profiles, network and content. The Jena framework [Carroll et al., 2004] supports the use of ontologies in OWL-DL with the rules reasoning (Jena’s Rules). The inferences on the ontology are made by an inference machine (reasoner) of this framework. In the SADP process this framework is used for consultations on the fuzzy ontology of the site. The Tool Obtain Information of Site Access (TOISA) aims at obtaining implicitly information on the user's access, without the perception or even the spent of time of the user. These information are used to update the profiles. The Tool Treat Profiles (TTP) creates and updates the user profiles, device, network and content, from information collected by TOISA. The Tool Personalize Site (TPS) is responsible for personalizing and adapting the contents based on the profiles. 98 IADIS International Conference e-Commerce 2009 4. SADP PROCESS The SADP process is divided into four activities: Develop Fuzzy Ontology, Obtain Information of Site Access, Treat Profiles and Personalize Site. Figure 1 shows, in the Structured Analysis and Design Technique (SADT) notation [Ross, 1977], the first activity of the process. In this activity the software engineer starts from the information about the site domain, stored in the site database, and others information such as sales reports and site access. Guided by fuzzy meta-ontology OWL-DL it develops a fuzzy ontology that represents the site knowledge formalization based on fuzzy logic. Site information are obtained from the data models of the products available on the site. Sales and access to the site reports, for example, are also sources of information that can be used in the construction of ontology. These and other details, such as the pertinence degrees of the relationships between the products, are analyzed by a software engineer on the fuzzy ontology modeling. The OWL-DL orients the software engineer in the modeling of products and their fuzzy relationships. Site domain Information Site Database Fuzzy Meta-Ontology OWL-DL 0.5 0.75 0.3 0.8 Develop Fuzzy Ontology Site sales and access reports Soft. Eng. 0.95 Site Domain Fuzzy Ontology Figure 1. SADP Process – Part 1 With the development of Fuzzy Ontology it is explicited and specified the following information from the site products: product repurchase potential (RepurchasePotential), amount of information that a user must know and consider before buying a product (ProductComplexity) and the fuzzy relationships between products (pertinence degree). After the site fuzzy ontology was built, the process SADP continues with the Obtain Information of Site Access, Treat Profiles and Personalize Site activities, as shown in Figure 2. These activities are repeated in a cycle in each access of the user to the site. In Activity Obtain Information of Site Access, first the software engineer installs TOISA on the site. A reference to a remote script that uses the AJAX technology is added to the pages of the site [Garrett; James, 2005]. The TOISA uses this script to obtain implicit information of user’s access. When a user accesses the site for the first time, it is assigned a unique identification (UserId) randomly created based on the millisecond of the moment of his access. This identification is stored in a cookie (file stored on the user's machine, which contains information exchanged between the server and the web browser) and in a database for use in subsequent accesses. In case the cookie option is not released in the user's browser, the identification is performed by IP address. Figure 2. SADP Process – Part 2 Through the UserId it can be executed the recovery and persistence of the user access information. Figure 3 presents an example of information collected from a user access to the site. 99 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 3. Information of the User Access to the Site In Activity Treat Profiles, the profiles are built based on information collected by TOISA. These profiles are stored in the database for queries and updates during the SADP process. For the user profile, the TTP first processes the information persisted by TOISA to determine the user's navigational behavior. This navigational behavior includes the following information: number of times the user accessed a particular product (NumberAccessProduct), total access time (TotalAccessTime), purchase execution (PurchaseExecution) and geographical location (city and state). These information are used to define a score for the content accessed (AccessScore) and bought by the user (PurchaseScore). This score is processed by the combination of the navigational behavior with the product complexity (ProductComplexity) and repurchase potential (RepurchasePotential). Figure 4 shows the algorithm used to generate scores from the user’s Registry of Access Information (RAI). The variables T1, T2, T3, T4 and T5 represent the access time, in seconds, to site products. These times can be set according to the portage and nature of the site contents. Obtain UserId; Create ProductList and initiate the variable with empty; For each RAI do If CodProduct is associated with UserId and CodProduct is not at ProductList then Add CodProduct in ProductList; EndIf EndFor For each CodProduct in ProductList do Retrieve ProductComplexity and RepurchasePotential of the product of the Fuzzy Ontology of the site Assign variables AccessScore=0 and PurchaseScore=0; For each RAI with CodProduct associated with UserId do If ProductList.CodProduct = RAI.CodProduct then Add 5 score to AccessScore; Retrieve TotalAccessTime and PurchaseExecution in RAI; If PurchaseExecution = 1 then Add 1 to PurchaseScore EndIf If ProductComplexity = 1 then Case: T1 < TotalAccessTime < T2 : Add 2 scores to AccessScore; T2 <= TotalAccessTime < T3 : Add 3 scores to AccessScore; TotalAccessTime >= T3 : Add 4 scores to AccessScore; EndCase EndIf If ProductComplexity = 2 then Case: T1 < TotalAccessTime < T2 : Add 1 score to AccessScore; T2 <= TotalAccessTime < T3 : Add 2 scores to AccessScore; T3 <= TotalAccessTime < T4 : Add 3 scores to AccessScore; TotalAccessTime >= T4 : Add 4 scores to AccessScore; EndCase EndIf If ProductComplexity = 3 then Case: T2 <= TotalAccessTime < T3 : Add 2 scores to AccessScore; T3 <= TotalAccessTime < T4 : Add 3 scores to AccessScore; T4 <= TotalAccessTime < T5 : Add 4 scores to AccessScore; TotalAccessTime >= T5 : Add 5 scores to AccessScore; EndCase EndIf EndIf EndFor Associate RepurchasePotential, AccessScore and PurchaseScore to CodProduct; EndFor Return ProductList. Figure 4. Algorithm for Score Attribution to the User Interests on the Site Products 100 IADIS International Conference e-Commerce 2009 Continuing in activity Treat Profiles, it is added new products to the user profile as the algorithm shown in Figure 5. From a product Pi of preference of a User U, new products recommendations are obtained by consulting the fuzzy ontology (OWL-DL), through the Inference Module on Fuzzy Ontology (IMFO) and the Jena framework. The fuzzy ontology already provides for the Related Products (Pr) their pertinence degrees μ and RepurchasePotential. The AccessScore of Pr is obtained by multiplying the AccessScore of Pi by pertinence degree μ of the relationship between Pi and Pr. The other profiles are created considering the features of each device category, network and content. These profiles are related to the user profile to define the personalization and adaptation of the different architectures according to each content type, device or network access. Once constructed, the profiles are persisted in a database using the Adaptive Object-Model architecture [Yoder et al., 2001]. The Adaptive Object-Model architecture uses the ObjectType pattern, which separates an entity of its type of entity and property of its type of property. The Adaptive Object-Model architecture turns flexible the dynamic inclusion of new features, following the continuing evolution of the profiles. Obtain User U and CodProduct Pi do Retrieve of Fuzzy Ontology the products related to CodProduct Pi, along with their pertinence degree μ and RepurchasePotential; Create ProductsRelatedList; For each Product Related Pr do AccessScore of Pr = AccessScore of Pi * µ; Add Pr in ProductsRelatedList; EndFor Return ListProductsRelated; Figure 5. Algorithm for Products Addiction in the User Profile In the Activity Personalize Site in the SADP process, from the profiles and user site access the Tool Personalize Site (TPS) customize a page architecture with products of the site and display it to the user. To do this, once a user is identified, the products that will compose the personalized page are determined. The priorities for recommended products selection are based on the scores defined in the profile of each user, according to his preferences. First it is recommended products with higher PurchaseScore which PurchaseScore are high and medium, followed by products with higher AccessScore. Once defined the products and their priorities for the composition of the page, the TPS uses the device profiles, network and content, and the framework UBICK, to adapt the content and send this adapted content to the user's device. For better understanding of the SADP process, it is presented next a case study that illustrates each step of this SAPD process. 5. CASE STUDY The SADP process has been evaluated in a site called MalibuFestas, with public access of its users. This is a site that sells products for parties in general, with access through the url: http://www.malibufestas.com.br. In the Activity Develop fuzzy ontology of the site domain, the fuzzy ontology was built from the information on the site contents stored in database and in other requirements to be attended with the personalization. It is shown in Figure 6 a passage of the MalibuFestas site fuzzy ontology modeling, with the fuzzy relationships between its contents. ISA isAdditional isSimilar Sixties Drinks TROPICAL_DRINK_MIXER 0.85 LUXURIOUS_DRINK_MIXER GlassesIce SNACKS_ TAGGER 0.9 FLASHING_BIG_GLASS 0.8 FLASHING_BEER_GLASS CHALICE Figure 6. Fuzzy Ontology Model of the MalibuFestas Site Contents 101 ISBN: 978-972-8924-89-8 © 2009 IADIS In this model, the classes Drinks and Glasses-Ice, Sixties subclasses, represent a product category of the site that has instances representing its products. The properties isSimilar and isAdditional are used to define the relationships between instances. For these fuzzy relationships, meta-tags of the Fuzzy Meta-Ontology ("fuz" prefix) are added to the OWL-DL. Figure 7 shows the specification in OWL-DL of the Figure 6 ontology, using these meta-tags. In (A), there are the information RepurchasePontential and ProductComplexity of the product Chalice, and their relationships defined by the isAddicional property. In (B), the isAddicional property is used to specify the pertinence degrees of these relationships, as between Chalice and Luxurious Drink Mixer (0.9). Figure 7. OWL-DL Fuzzy Ontology In the Activity Obtain Information of Site Access, once installed, the Tool Obtain Information of Site Access (TOISA) will collect the information from the user’s access to the site that will be used in developing profiles. As an example, the following information were obtained for a particular user: UserId 1225823398758, date and access time 02/25/2009 - 15:32, IP adress 192.168.0.169, an access to the home page with 32 seconds of total access time and two accesses to the product Chalice (code 854) with total access time of 55 seconds. Following, in the Activity Treat Profiles, the TTP processes the information of the site access, based on Figure 4 algorithm, to generate the scores of the user's preferences. For example, for UserId 1225823398758, 8 preference points were generated for the product Chalice. To obtain this score in the algorithm of Figure 4, two variables refering to the user's navigational behavior were considered: NumberAccessProduct and TotalAccessTime. The process was applied from January 28 to February 28 of 2009. Figure 8 shows these variables for 16 users and 4 products of the site. Product C halice User User User User User User 1225823398758 1234392127621 1235745241193 1234390157538 1235666910334 1235174501321 Product Snacks Tagge r User User User User 1234373845964 1233687178811 1234523633210 1233667736583 Numbe r Acce ss Product 2 6 3 1 1 1 Total Acce ss Time 19 12 14 10 22 9 Purchase Exe cution Numbe r Acce ss Product 4 3 1 1 Total Acce ss Time 13 17 13 18 Purchase Exe cution Yes Yes No Yes No No Yes Yes No No Product Luxurious Drink Mixe r User User User User User User 1234890385395 1234390076463 1233742720626 1235175547377 1234829252612 1232654602442 Product Flashing Small Glass User User User User 1234373845133 1233687174123 1234524633998 1233667737789 Numbe r Acce ss Product 4 3 4 2 2 1 Total Acce ss Time 11 14 13 18 22 8 Purchase Exe cution Numbe r Acce ss Product 4 2 2 1 Total Acce ss Time 39 16 22 18 Purchase Exe cution Yes Yes Yes No No No Yes Yes No No Figure 8. Users Navigational Behavior of the Site MalibuFestas Analyzing these behaviors, it was verified that the users that accessed a product more often had a greater interest in it, accomplishing their purchases. The users that accessed only once the products had 16% of purchase accomplishment. Users that accessed twice the same product had 40% of purchase execution, and users that accessed three or more times the same products had 85% purchase execution. The purchase accomplishments were higher for larger numbers of access. Another fact is that not always the greatest time 102 IADIS International Conference e-Commerce 2009 of access to certain product corresponded to the accomplishment of the purchase. Thus, the information about NumberAccessProduct has greater relevance in the calculation of the scores of the user's interests. Exceptionally, in some cases where the user is already a customer of the site, as the user 1234390157538, the purchase can occur with low total access time and with little number of accesses to the product. Once defined the access score for the Chalice product, the TTP consults the site fuzzy ontology and retrieves products related to these products. Then, as the Figure 5 algorithm, the list of related products is produced. In this case, this list has the Luxurious Drink Mixer (μ = 0.9 in relation to the Chalice), which has the AcessScore of 7.2 (0.9 * 8 points) and the Snacks Tagger (μ = 0.8 in relation to the Chalice), which score is equal to 6.4 (0.8 * 5 points). Besides the user profile, the device, network and content profiles are developed based on information of user access. Finally, In the Activity Personalize Site, a list of the most relevant contents to the user is produced (RecommendationList) and, dynamically, the architecture of the page to be sent to the user is built. For the page construction, the other profiles, device, network and content related to the user are consulted, for the adaptation of the contents. Based on the profiles, the UBICK framework makes the adaptation of the custom content. Thus, the architecture of the site is built dynamically according to the profiles. For example, Figure 9 shows the architecture of the page built for the user with their recommendation. Figure 9. Recommendation List of the MalibuFestas Site 6. EVALUATION OF THE SADP PROCESS The evaluation has been done for the results of the SADP process aiming to calculate the rate of hits of its recommendations. This rate was calculated based on the users who received recommendations and performed purchases at the MalibuFestas site. Based on the user’s scores about the products, obtained by the algorithm of Figure 4, an analization of the users was made from March 03 to March 31 of 2009. The calculus of the recommendation rate of hits considers the purchases made and the products recommended to the user. Table 1 lists the products recommended and purchase accomplishment for some of the users examined. Table 1. Relationship between Recommended Products and their Execution Purchasing User 1225823398758 1237981758035 1237940820384 1237825898480 1237212314641 1237826763230 ... Total Products Recommended 12 12 12 12 12 12 ... Purchases of Recommendation 6 1 10 3 0 4 ... Based on this analyse, it was shown that 85% of the products bought were present on the list of products recommended by SADP. Only 15% of the purchases were not included in the users profiles recommended by the SADP. It was also verified that 90% of the products of the profiles of users that were purchased received scores higher than products not purchased. Thus, with the analysis it can be concluded that the personalization process offered presents satisfactory results, with increase in the sales of MalibuFesta site. 103 ISBN: 978-972-8924-89-8 © 2009 IADIS 7. RELATED WORK Some works about content personalization make use of semantic information (Ontologies) to build the profiles or the personalized pages, such as Mobasher (2005) and Palazzo et al. (2006). Unlike those, in the SADP process the fuzzy ontology is used to define information with an uncertainty degree, as the pertinence degree, making more accurately the priorities of the recommendations. In Gotardo (2008) works the discovery of the user preferences is based on the visited contents, but organized in groups. In the SADP process the profiles allow the treatment of each user with his specific preferences. Other works, based on data mining, as in Aroyo (2006), use the similarity relationship between the site contents to recommend new products to the user. However, these similarities consider only the complete relations between the products. Another point to note related to the correlated works presented is that on SADP process the page architecture is driven by personalization, considering the content adaptation to access devices. In the process, the profiles are kept updated at each new user's access, keeping up with the continuous preferences changes. 8. CONCLUSION This paper presented a process of content personalization (SADP), which examines aspects of the user's navigational behavior and extends his preferences based on fuzzy relationships between products modeled with fuzzy ontology. The SADP process also considers the adaptation of the content of the site to the access device. The user, device, network and content profiles, along with a framework for adaptation of content, are used to construct dynamically the architecture of the site. It was presented a case study that illustrates how the site architecture can be directed by personalization, presenting to the user the site content personalized and adapted to his device and access network. The project continues with new studies that include more detailed analysis on the effectiveness of the process and a study on the relevance of the user's navigational behavior in the execution of purchases. Another work in progress aim at improving the way of scoring the user preferences. REFERENCES Aroyo, L. et al, 2006. Ontology-based personalization in user adaptive systems. 2nd International Workshop on Web Personalization, Recommender Systems and Intelligent User Interfaces. Dublin, Ireland. Berners-Lee. et al, 2001. The Semantic Web. Scientific American. USA. pp. 284(5) : 34–43. Carroll, J. et al, 2004. Implementing the semantic web recommendations. In International World Wide Web Conference. New York, NY, USA, pp 74–83 Garrett, and Jesse James. 2005. AJAX: A New Approach to Web Applications. Adaptive Path. San Francisco, USA. Gotardo, R. et al, 2008. An Approach to Recommender System Applying Usage Mining to Predict Users Interests. 15th International Conference on Systems, Signals and Image Processing. Bratislava-Slovak Republic, v. 1. p. 1-4. Gruber T. 1993. A Translation Approach to Portable Ontology Specifications. Knowledge Acquisition. Boston. v. 5, n. 2, pp. 199-220, 1993. Mobascher B. and Daí, H., 2005. Integrating Semantic Knowledge with Web Usage Mining for Personalization. In Web Mining: Applications and Techniques. San Jose, California. Palazzo J.M. and Rigo S.J., 2006. Aquisição automática de classes de usuários integrando mineração de uso da Web e ontologias. Workshop em Algoritmos e Aplicações de Mineração de Dados. Florianópolis, SC, Brazil. Ross, D., 1977. Structured Analysis: A Language for Communicating Ideas. IEEE Transactions on Software Engineering 3(1), Special Issue on Requirements Analysis. pp 16-34. Santana, L. H. Z. et al, 2007. Adaptação de Páginas Web para Dispositivos Móveis. Simpósio Brasileiro de Sistemas Multimídia e Web. Gramado, RS, Brazil. v. 1. p. 1-8. Yaguinuma, C. et al, 2007. Meta-ontologia Difusa para representação de Informações Imprecisas em Ontologias. In: II Worshop on Ontologies and Metamodeling in Software and Data Engineering. João Pessoa, PB. v. 1. p. 57-67. Yoder, J. et al, 2001. Architecture and Design of Adaptive Object-Models. Proceedings of the ACM SIGPLAN Conference on Object Oriented Programming, Systems, Languages and Applications (OOPSLA 2001). Tampa, Florida, USA. 104 IADIS International Conference e-Commerce 2009 A CONTENT ANALYSIS OF WEB-SITE QUALITY OF ONLINE AUCTION SELLERS Fen-Hui Lin and Chiu-Chu Hwang National Sun, Yat-sen University 60, Lien-Hai Rd., Kaohsiung, Taiwan ABSTRACT Maintaining good web-site quality is essential to gain the trust of Internet potential customers. Thus, it is of great interest to examine how the highly successful e-auction sellers present their products and project their credibility through their web-site design and layout. This study seeks to investigate the web-site quality of successful e-auction sellers in Taiwan. We have chosen the e-auction sellers with the highest customer ratings on Yahoo!Kimo on-line auction in Taiwan and we have performed a content analysis on a sample of 392 chosen sellers based on fourcriteria: graphics, structure, content and social-cue design. The web-site design, mostly graphics and pictures are used. However, the quality of the pictures are not as good and refined as those of the other on-line stores. In their structural design, most of the web-sites presented all the necessary information. In the content, more than 80% of the web-sites displayed their special logos to re-enforce their company identity. The most listed product information includes prices, components, characteristics, quarantee, quality, etc. For the social-cue design, most of the web-sites listed their phone numbers or Internet addresses. In addition, about half of the web-sites provided the street addresses of the physical stores and their operating hours. KEYWORDS On-Line Auction, Web-Site Quality, Trust, Content Analysis. 1. INTRODUCTION The e-auction selling system has created a booming C-to-C Internet business model. Because of the low initial investments needed in order to enter the business, e-auction has attracted many small business start-ups to sell on-line. There is no need to maintain any physical stores. There is no store space rental or sales personnel to pay for. There is likewise little need to maintain a huge stocks inventory. All these have resulted in the following phenomena: z Small business start-ups have developed into small or medium scale enterprises or even international business undertakings, such as ”Tokyo Fashion 1 ” that was started in 2004 by a junior undergraduate student named ”Mayuki.” (This is the username displayed on the Yahoo!Kimo e-auction account). In two years’ time , Tokyo Fashion had realized the highest revenues and received the highest postive auction ratings among more than seventeen hundred women’s clothing e-auction sellers on Yahoo!Kimo. Since then, it has maintained its place on the top of the ratings. Its annual revenues has amounted to more than four million dollars. Tokyo Fashion has also started another e-auction business on the taobao.com2 in China (Newsweek, 2007). z Even individuals such as single parents, home-bound entrepreneurs or indigenous people, can sell handmade arts and crafts through the Internet. The e-auction system provides a convenient and low cost platform for these people to promote and sell their products in order to increase their earnings. Moreover, it also helps the indigenous people to preserve and promote their particular culture and heritage by providing an outlet for their arts and crafts. However, because of the popularity of the e-auction system, more and more sellers now crowd the Internet and offer an enormous choice of merchandise. To grab the attention of Internet shoppers and entice 1 2 http://tw.user.bid.yahoo.com/tw/booth/mayki0920 http://www.taobao.com 105 ISBN: 978-972-8924-89-8 © 2009 IADIS them to purchase a certain product, it has become very critical for e-auction sellers to be highly competitive in the Internet business environment. Because of the fast speed and great convenience involved in an Internet search, the task of searching for information and comparing prices for particular items has also become easier. It has been proved that the primary reason why people purchase merchandise on the Internet is because of the lower prices of items when compared with buying them from a physical store. A general perception among buyers of prevailing lower prices on the Internet also causes Internet sellers to further lower their prices so that their profit margin is further squeezed. However, although the e-auction Internet business is highly competitive and only has low profit margins, many e-auction sellers achieve high volumes of sales and revenues. Therefore, discovering the operating characteristics of successful e-auction sellers can only help other sellers to improve their own business efficiency and performance. This study aims to investigate the more successful e-auction sellers based on the quality of their web-sites by using the content analysis methodology. For a good number of e-auction sellers, the trust issue is an important consideration for on-line buyers when making their purchases. (Gefen, Benbasat and Pavlou, 2008). The web-page has been the most essential communication channel between sellers and potential buyers. The layout of a web-page is the primary vehicle where a seller can earn the trust of potential buyers and entice them to buy. (Lowry, Vance, Moody, Beckman and Read, 2008; Cyr, 2008). The Internet shoppers put a lot of importance on web-site brand and credibility. The sellers have to minimize any suggestions of a transaction risk for potential customers by coming out with well-designed web-pages. Wang and Emurian (2005) suggested a scheme with four web-site design dimensions in order to gain the trust of potential Internet customers: graphic design, structural design, content design and social-cue design. 1.1 The Graphic Design Dimension Web-site graphics dictate the first image that Internet shoppers get about a product. The overall composition of the graphics and web-page colors also affect the feeling of trustworthiness of a web-site (Kim and Moon, 1998). Moreover, a sense of quality and professionalism would likewise further raise the degree of trust by the customers. Basso, Goldberg, Greenspan and Weomer (2001) point out that attractive graphics quickly grab the prospective buyer’s attention as well as demonstrate the capability and professionalism of the online sellers. The pictures used should be of high quality and be carefully chosen. As much as possible, line drawings should not be used. In order to achieve a good visual effect, the web-page design should integrate the layout, text fonts and colors, presenting an almost true to life picure of the product. All these efforts can only improve the credibility of the web-site, in general, and the product, in particular. 1.2 The Structure Design Dimension Wang and Emurian (2005) suggested that the web-site’s structure can also be examined from two aspects of information presentation: overall organization and accessibility. The overall organization indicates the ease of use and easy readability of the web-page inter-face and structure. A web-site presented with simplicity and consistency can effectively raise the feeling of trust with the customers. (Neilsen, 1998; p 107). Accessibility is another important factor which includes the proper arrangement of the hypertext and figures, the sequence of the web pages. The use of white space, or groupings can further improve readability and overall credibility (Zhang, von Dran, Small and Barcello, 1999). 1.3 The Content Design Structure Content design is about how the web-site presents itself overall. In order to enhance the perception of trust, four aspects should be considered. The first is the brand image of the web-site that includes the company logo, business background, the range of operations and contact details, Egger (2001) suggested that a conspicuous logo with slogans together with a strong emphasis on the company merchandise can very well strengthen the company image and encourage inquiries from customers. The second aspect is customer relationship management. The web-site should contain the following information: the company capability, provisions for security and privacy, finance and legal responsibility, etc. Many customers are usually very concerned about security and privacy policies, the tracking of 106 IADIS International Conference e-Commerce 2009 merchandise delivery, packaging and delivery services, the refund policy and contract limitations and so on. Such information should be prominently displayed on the web-page (Cheskin, and Sapient, 1999; Egger, 2001; Neilsen, 1999). The third is to demonstrate the accountability of the company through the web-site. The company logo or text descriptions can effectively demonstrate the professionalism of the e-commerce web-site. Hu, Lin and Zhang (2001) suggested five approaches that can underscore the company’s stability: protection of customer privacy, guarantee of the transaction’s security, testimonies of customer satisfaction, statement of of credibility and product or transaction warranty. The fourth aspect is to provide full product information accurately and in a timely fashion. Embedding the brand name or company name in the web-site address also increases the feeling of trust by customers (Egger, 2001; Nielsen, 1990). This is similar to the advertising effect that is brought about when providing delicate and useful information in advertisements that can provide assurance to the customers (Hunt, 1976). Therefore, a given brand can further differentiate itself from other competitors so that it becomes more attractive and the customers’ understanding of the product can be enhanced. For the surface content of the web-site, there are two essential parts, the first is title and the second is the provision of product information. Beltramini and Blasko (1986) categorized six types of titles shown on traditional advertisements. These include familiar sayings, contrasting statements, news/information, shock effect, questions and arousing reader’s curiosity. After studying award-winning advertisements in the United States, they found that 55.9% belonged to the first two categories (familiar sayings and contrasting statements). Considering the product information presentation, Resnik and Stern (1977) generated fourteen product characteristics making use of criteria that have been adopted by several other researchers for their own researches (Abernethy and Franke, 1996). Many researches had proved that when sufficient and applicable information are placed within the advertisements, the better the customers are able to make purchasing decisions. It has also been shown that the advertisements for durable and expensive products contain richer product information. Comparatively, the advertisements for food, household goods and appliances, personal hygiene products, etc. contain lesser amount of information (Resnik and Stern, 1977; Stern, Krugman, and Resnik, 1981). 1.4 The Social-cue Design Dimension The social-cue design in a web-site can make use of various communication media to reflect the current social milieu or to improve on the face-to-face interaction. Riegelsberger and Sasse (2002) suggested that the lack of inter-personal contacts on the Internet can create a gap between customers and retailers so that the customers’ trust for a product may wane. To overcome this problem, the web-site can use pictures of company operations to display its operating details. Steinbruck, Schaumburg, Duda, and Kruger (2002) performed an experiment and found that projection and connection with customers can be generated via a social-cue design. Posting pictures or photos that show company workers is a simple and effective way to further strengthen customers’ trust. Basso et al (2001) found that the media interface that uses a more interactive design can indirectly influence customers to share information about the company and its product or to purchase again. Generally, customers spend an average of five minutes on a web-site. Providing facilities for simultaneous communications, such as instant messenger or Internet telephone connections allow customers reach sellers easily, thus, minimizing any misunderstandings in the process. This also builds a good impression for the products and services provided through the website starting with the customer’s first visit This is an exploratory research using content analysis to find out more about web-site design of those e-auction sellers who received the highest rankings based on the four dimensions of Design. The following section is a description of the content analysis research methodology while the third section presents the empirical results. The fourth section presents the discussions and managerial implications; and finally, the conclusions of this research. 107 ISBN: 978-972-8924-89-8 © 2009 IADIS 2. THE CONTENT ANALYSIS RESEARCH METHODOLOGY Based on the criteria identified for the e-auction web pages, we conducted a content analysis to investigate the most successful e-auction sellers on Yahoo!Kimo in Taiwan, i.e., web-page design, lay-outs and contents. Content analysis has been deemed to be a proper approach when examining the messages contained in the texts or graphics with systematic, objective and quantitative characteristics (Kerlinger and Lee, 2000). 2.1 Research Subjects and Sampling Because Yahoo!Kimo has the highest Internet traffic portal in Taiwan and has the highest transaction volume for e-auctions when compared with other e-auction platforms, the research subjects chosen are those eauction sellers of that have been in the business for a given period of time and who have reached the highest ratings/values in its merchandise category in Yahoo!Kimo’s twenty merchandise categories. The auction accounts were chosen for content analysis according to the following three criteria: z eighty percent of the merchandises displayed by the auction account belongs to the same category; z the auction account is on the top twenty-five of the reputation ratings in the category that it belongs to; and z the reputation rating is higher than 1000. For each e-auction account, four types of web pages were collected, including (1) the virtual store layout, (2) ”about me,” (3) reputation ratings and comments, and (4) merchandise descriptions and pictures. Except for the fourth type that can number from one web page to more than ten web pages, the other three types each can be contained in only one web page. For the forth type of web page, we chose the first listed merchandise shown on the first page of the auction account that is linked to a web page that has photos of the merchandise associated with the introductory texts or item series numbers. The sample collection was performed from 2007, January2007 to February 2008. Finally, a sample of size 392 was obtained from eighteen categories of Yahoo!Kimo. The categories, “automobiles” and “travel, estate and service” were not included because none of their auction accounts garnered a rating value higher than 1000. Table 1 lists the positive ratings, negative ratings and ratios of positive ratings to negative ratings for the eighteen categories. The average positive rating is 6281.5 (the bottom row of Table 3). The top three positive ratings are “Women’s clothings and accessories” (rating = 25251.5), “Health and personal care” (rating = 11026.7), and “Women’s handbags and shoes” (rating = 10804.8). For the negative ratings, the average is 10.7, while the top three categories are in the same in the order as the positive ratings. The average ratio is 99.89% with half of the categories having higher ratios. Moreover, the top three categories in terms of ratio value are “Food and regional specialties,” “Mom and Baby” and “Sports and outdoors.” In general, the negative ratings are relatively small when compared with the positive ratings. Moreover, for the two handling approaches of the negative ratings, the total numbers for “benevolence and concern” and “inappropriate responses” are 593 and 563, respectively. The average numbers are 1.51 and 1.44. 2.2 Pretest, Reliability and Coding After forming the first draft of the coding scheme, a pre-test was performed by, the author and two MBA graduate students. A sample of thirty auction accounts included in the research sample were chosen to test the drafted coding scheme. After some minor modifications after the pretest, the final coding sheet for the formal analysis was obtained. The formal coding process (Krippendorff, 2004; Son, Tu and Benbasat, 2007) was then done. The authors held a training session to guide the coders to fully understand the definitions and boundaries of eachcriterion. Then, one pilot test was performed to test the consistency among three coders. One hundred and twenty accounts were drawn from the full sample of 392. The three coders performed the coding procedure separately for these 120 accounts. The reliabilities computed using Holsti’s formula (Holsti, 1969) for each criteria that is above 0.9 (Wimmer and Dominick, 1994). The formal coding period was made in May, 2007. 108 IADIS International Conference e-Commerce 2009 Table 1. The Positive Ratings, Negative Ratings and Ratios for Various Product Categories Merchandise Categories (sample size) Art collections (15) Books and magazines (17) Cell Phones (23) Computers (25) Digital cameras, video cameras and web cams (18) Electronics (25) Food and Regional specialties (22) Health and Personal Care (25) Home and Garden (25) Jewelry and watches (25) Men’s clothing and accessories (25) Mom and baby (25) Music and DVDs (13) Office (10) Sports and outdoors (24) Toys and games (25) Women’s clothing and accessories (25) Women’s handbags and shoes (25) Average (total sample size = 392) Mean of positive ratings (P) 2140.3 2966.6 3533.4 5914.9 4733.3 2762.9 1968.0 11026.7 3751.7 4241.8 8002.6 4490.7 3843.3 2545.2 4498.5 3219.1 25251.5 10804.8 6281.5 Mean of negative ratings (N) 1.87 1.41 3.57 4.52 4 2.1 0.8 17.3 4.6 2 5.2 1.9 2.9 2.5 2.0 2.7 97.2 16.8 10.7 3. RESEARCH FINDINGS AND RESULTS 3.1 Graphic Design Dimension There were only two web sites that did not use any graphics. Table 2 summarizes the empirical data on the graphic quality and graphic content. For the graphic quality, over 60% of the web sites show graphics with amateurish image processing. However, there are 34% (about one-third) of the web-sites posting graphics with professional quality. These can be expected to attract customers’ attention and enhance the sellers’ image as to their professionalism and competence. As to the graphic content, about 90% of the web-sites show pictures containing only one product with only two or three lines of graphics. Only very few of the web-sites show pictures containing several products. It could be because this would fail to focus on product characteristics. According to Table 2, there are 40% of web-sites showing ”moving” pictures. 3.2 Structural Design Table 3 lists the counts of the three approaches. Seventy-eight per cent of the web-sites have satisfactory accessibility of information. Fifty-nine per cent of the web-sites have the proper page design, such as appropriate white spaces and margins, groupings and visual density. A third of the web-sites provide functions for navigation reinforcement, such as prompts, guides, tutorials and instructions for transactions. 109 ISBN: 978-972-8924-89-8 © 2009 IADIS Table 2. Summary of the Graphic Design Dimension Graphic Quality Frequency (Percentage) Amateurish quality 241 (61%) Professional quality 135 (34%) Combination 14 (4%) No graphics 2 (.5%) Total 392 Graphic Content Picture of Product only Product picture with product information Multiple-product picture Multiple product pictures with product information None pictures Total Frequency (Percentage) ”Moving” Picture Frequency (Percentage) 270 (69%) Yes 156 (40%) 81 (21%) No 236 (60%) 19 (5%) 11 (3%) 2 (1%) 392 Total 392 Table 3. Summary of the Structural Design Dimension Structural design Frequency(F) Information accessibility Page design Navigation reinforcement Total counts (TC) 307 235 124 Web site average(= F / 392) 78% 59% 32% 666 3.3 Content Design Dimension There are three measures of web site contents: logos, product titles and product information. The empirical results are summarized in Table 4. For the first item as logo posting, most of the web-sites do show their unique logos for 83% of the sample size. For the second item, auction title, on the top two lists are “informativeness” and “benefits” with almost 90% of the sample. This indicates that a direct description of the product, providing its benefits, still are the most used strategy by sellers to promote their merchandise. For product information, the total count is 1290. With the s392 ample size,, the average information provided by each web-site is 3.3 items. The top three on the list are “price and value,” “components and contents” and “product characteristics,” representing almost 70% of the total. The three items are direct descriptions of the merchandise. This indicates that product information is the most important information provided by the e-auction sellers. 3.4 Social-cue Design This criterion is used to record the contact methods provided on the web-site. The empirical results of the three items are listed in Table 5. Most of the web-sites provide phone numbers or Internet contact information for the customers to use to communicate with the seller. These include mobile phone numbers, customer service phone numbers, email address, or other synchronous communication media such as MSN, Yahoo!Messenger, ICQ, etc. Half of the web- sites provide their operating times when the sellers or its operators are on-line and are able to deal with on-line business-related activities or answer customer questions. For the third item, 27% of the web sites show a map and the street address of the store location that would imply that these e-auction sellers also run a physical store or own a company. 110 IADIS International Conference e-Commerce 2009 Table 4. Summary of the Content Design Dimension Logos frequency Yes No percentage 325 67 83% 17% 392 frequency percentage Informativeness Benefits 232 116 59% 30% Surprise Curiosity Questioning Others 25 18 1 0 6% 5% .03% Total 392 Total Auction title Product information frequency percentage Price and value Component and content 392 269 100% 69% Characteristics Guarantee and service Quality Promotion activities Taste Safety and security Novelty Independent study 225 139 57% 35% 121 60 31% 15% 42 22 14 6 11% 6% 4% 2% Total 1290 Table 5. Summary of Social-cue Design Dimension Contact information Frequency (F) Percentage (F/392) Phone number or Internet communication media Operating time Map and address of store location 359 91% 195 106 50% 27% Total 560 4. CONCLUSIONS This study aims to investigate the web-site quality of successful e-auction sellers in Taiwan. The samples in this study are the sellers with the highest ratings in the Yahoo!Kimo on-line auction. They are long term auction sellers. The research results show that these web-sites have their own particular styles and images. All sought to create a professional image so to enhance the trust of potential customers. The graphic design, graphics and pictures were used to display the merchandise. However, the quality of some of the pictures was not as good and refined as those of the other on-line stores. This might be because most of the e-auction sellers are small businesses or a one-person business, without the special skills for photography or did not have the funds needed for the extra investment for this purpose. For the structure design, most of the web-sites included made it easy to access additional information The second focus would be the web-page layout with consistent styles to integrate the pictures and text. For the content design, more than 80% of the web-sites show their company logos to re-enforce their company identity. The auction titles are used mainly for infomation or product benefits. The most listed product information contains prices, product components, characteristics, wuarantee, quality, etc. For the social-cue design, 91% of the web-sites list the phone numbers or Internet address. In addition, about half of the web-sites provide the address of the physical store and its operating hours in order to create the image that Internet customers can trust their businesses. The sellers make sure to answer the phone during their business hours. ACKNOWLEDGEMENT The authors would like to thank the national science council of Taiwan (NSC 97-2410-H-110-030) for the support of this study. 111 ISBN: 978-972-8924-89-8 © 2009 IADIS REFERENCES Abernethy, A.M. and Franke, G.R., 1996. The information content of advertising: A Meta-Analysis. Journal of Advertising Research, Vol. 25, No. 2, pp 1-17. Basso, A., Goldberg, D., Greenspan, S. and Weimer, D., 2001. First impressions: Emotional and cognitive factors underlying judgments of trust e-commerce. In Proceedings of the 3rd ACM Conference on Electronic Commerce, pp 137-143. Tampa, FL, USA. Beltramini, R.F., and Blasko, V.J., 1998. An Analysis of Award-winning Advertising Headlines. Journal of Advertising Research, Vol. 26, No. 2, pp 48-52. Cheskin Research and Studio Archetype/Sapient., 1999. Ecommerce trust study. Retrieved November 7, 2006. Available: http://www.cheskin.com/p/ar.asp?mlid=7andarid=40andart=0. Cyr, D., 2008. Modeling Web Site Design Across Cultures: Relationships to Trust, Satisfaction, and E-Loyalty, Journal of Management Information Systems, Vol. 24, No.4, pp 47-72. Day, R., 1980. Research Perspectives on Consumer Complaining Behavior, Chicago: American Marketing Association. Egger, F.N., 2001. Affective design of e-commerce user interface: How to maximize perceived trustworthiness. In Proceedings of the International Conference on Affective Human Factors Design. London: Asean Academic Press. Gefen, D., Benbasat, I. and Pavlou, P.A., 2008. A Research Agenda for Trust in Online Environment, Journal of Management Information Systems, Vol. 24, No. 4, pp 275-286. Hosti, O., 1969. Content Analysis for the Social Sciences and Humanities. Addison-Wesley, Reading, MA. Hu, X., Lin, Z., and Zhang, H., 2001. Myth or reality: Effect of trust-promoting seals in electronic markets. Presented at WITS 2001. New Orleans, LA, USA. Hunt, S.D., 1976. Informational vs. Persuasive Advertising: An Appraisal. Journal of Advertising, Vol. 5, pp 5-8. Kerlinger, F.N. and Lee, H.B., 2000. Foundations of Behavioral Research, Fourth edition. Orlando, FL: Harcourt College Publishers. Kim, J. and Moon, J.Y., 1998. Designing towards emotional usability in customer interfaces － trustworthiness of cyberbanking system interfaces. Interacting with Computers, Vol. 2, No. 1, pp 1-29. Krippendorff, K., 2004. Content Analysis: An Introduction to Its Methodology. Thousand Oaks, CA: SAGE Publications. Lowry, P.B., Vance, A., Moody, G., Beckman B., and Read A., 2008. Explaining and Predicting the Impact of Branding Alliances and Web Site Quality on Intial Consumer Trust of E-Commerce Web Sites, Journal of Management Information Systems, Vol. 24, No. 4, pp 199-224. McKnight, H.D., Choudhury, V. and Kacmar, C., 2002. Developing and Validating Trust Measures for e-Commerce: An Integrative Typology, Information Systems Research, Vol. 13, No. 3, pp 334-359. Neilsen, J., 1999. Trust or bust: Communicating trustworthiness in web design. Jacob Nielsen’s Alertbox. Available: http://www.useit.com/alertbox/990307.html. Resnik, A. and Stern, B.L., 1977. An Analysis of Information Content in Television Advertising. Journal of Marketing, Vol. 41, No. 1, pp 50-53. Riegelsberger, J. and Sasse, M.A., 2001. Trustbuilders and trustbusters: The role of trust cues in interfaces to e-commerce applications. 1st IFIP Conference on E-commerce, E-business, E-government.Available: http://www.cs.ucl.ac.uk/staff/J.Riegelsberger/trustbuilders_and_trustbusters.htm. Taylor, R. E., 1999. A Six-Segment Message Strategy Wheel. Journal of Advertising Research, Vol. 39, No. 6, pp 7-17. Son, J.Y., Tu, L. and Benbasat I., 2006. A Descriptive Content Analysis of Trust-Building Measures in B2B Electronic Marketplaces, Communications of the Association for Information Systems, Vol. 18, No. 99-128. Steinbruck, U., Schaumburg, H., Duda, S. and Kruger, T., 2002. A picture says more than a thousand words – photographs as trust builders in e-commerce websites. In Conference Extended Abstracts on Human Factors in Computer Systems. Minneapolis, MN, USA. Stern, B.L., Kurgman, D.M. and Resnik, A., 1981. Magazine advertising: an analysis of its information content. Journal of Advertising Research, Vol. 21, No. 2, pp 39-44. Tokyo Fashion, Newsweek, Vol. 1009, March 2007. (assessed at Nov. 26, 2008) http://www.businessweekly.com.tw/webfineprint.php?id=28196. Wang, Y.D. and Emurian, H.H., 2005. An overview of online trust: Concepts, elements, and implications. Computers in Human Behavior, Vol. 21, No. 1, pp 105-125. Wimmer, R.D., 1994. Mass media research: an introduction. 4th ed. Belmont, California: Wadsworth. Zhang, P., von Dran, G.M., Small, R.V. and Barcellos, S., 1999. Websites that satisfy users: A theoretical framework for web user interface design and evaluation. IEEE Proceedings of the 32nd Hawaii International Conference on System Sciences, 1-8. Available: http://ieeexplore.ieee.org/xpl/tocresult.jsp?isnumber=16782andisYear=1999 112 IADIS International Conference e-Commerce 2009 GUIDELINES TO THE DEVELOPMENT OF AN E-COMMERCE PLATFORM FOR CUSTOMIZED GARMENTS Ribeiro, L. Master in Fashion Design, Textile Science and Technology Department, University of Beira Interior Pólo I – Rua Marquês d'Ávila e Bolama, 6201-001 Covilhã, Portugal Tel: +351275319700, Fax: +351275319768 Duarte, P. Management and Economics Department, Unit Research NECE, University of Beira Interior Pólo IV – Edifício Ernesto Cruz, 6200-209 Covilhã, Portugal Tel: +351275319600, Fax: +351275319601 Miguel, R. Textile Science and Technology Department, Textile & Paper Materials Research Unit, University of Beira Interior Pólo I – Rua Marquês d'Ávila e Bolama, 6201-001 Covilhã, Portugal Tel: +351275319700, Fax: +351275319768 ABSTRACT The world is changing and consumer behaviour with it. Today’s consumers expect to buy high quality customized products at low prices with fast delivery. This can be accomplished with Internet technology. Fashion design industry should understand the desires of these new consumers and develop e-commerce solutions in order to give a better response to the e-consumer needs and demand. Customization emerges as a form of cyber couture, the industry response to consumer desire to personalize their style. This paper aims to present the theoretical basis for a consumer oriented ecommerce platform, based on market research study. A web-based survey was conducted to understand online consumer buyer behaviour for customized fashion products. The results reinforce previous findings on the willingness of consumers to participate in the creation of personalized products and further extend literature knowledge on online consumer's behaviour in the fashion market. The result is a set of guidelines for the development and implementation of an ecommerce platform for customized fashion products. KEYWORDS E-commerce, Mass Customization, Fashion design, Consumer behaviour. 1. INTRODUCTION 1.1 Customization Current economy is highly competitive and competition increases continuously grounded in the globalization of markets. The changes and technological innovations are faster than ever before, providing a broad variety of consumption alternatives and stimulate news demands. This new “technological consumer” demands from fashion industry more quality, variety, exclusiveness and personalization. In the long run it is expected that mass production in textiles products will evolve for mass personalization – mass customization, where the consumer takes an active role is the process of fashion and garments development. The consumer needs to make a difference, the desire to distinguish oneself from others, the need to acquire unique products, with specific characteristics, that reflect its personality or lifestyle precedes the need of customization. One possible way to answer customer’s expectations is providing customized products by 113 ISBN: 978-972-8924-89-8 © 2009 IADIS the consumer himself, or at least give him the opportunity to play an active part in the definition of the final characteristics of a piece of clothing. Current customization applications are limited to a selection of some attributes, such as: shape, size, colours and components. Thus, the next level of mass customization should rely on technologic tools such as patterns visualization and the application of those in a 3D model to provide the consumer with the freedom and liberty to create their own clothes. The terms “customization” and “mass customization” lacks a precise definition since several authors focus different points of interest, and are still unknown for the majority of the population. The introduction in the European market is still recent and a little diffused (Piller, 2004). Sometimes is mixed-up with the concept of personalization, with some authors mixing the concept of personalization inside customization (Bae, May-Plumlee, 2005). 1.2 Customized Clothing and Online Commerce In the universe of customization consumers should be regarded as partners in the creation of value, which implies a new model of relationship between consumers and companies, based on mutual understanding. Some authors (Pine, Gilmore, 1999; Anderson et al., 2002; Piller, 2004) recognize the consumer’s disappointment with the products available in common stores, nevertheless they don’t want to just have a lot of choice, but the ability to choose among products that they really like. Looking into online customization of products it is necessary to be aware that it benefits with the use of technologies to incorporate images and videos to allow consumers to choose colours, shapes and styles desired, and combine them to create the desired product. Even though, the offer of those attributes should be addressed careful, because having them all available leads to consumer confusion and incapacity of choose and combine. So it is important to define a platform model that allows consumers to choose the preferred style of product and look at images of the final product. This way the consumer can have a more clear perception of the product. The adoption of a support system with these characteristics, provide the consumer with the ability to compare, choose and obtaining the best combination according to their selections from virtual images. In addition, the composition of images allows the consumer to see several combinations and avoid time waste choosing among a huge number of options (Lyengar, Lepper, 2000; Li. et al. 2000; Ma et al., 2007). The majority of the consumers don’t feel secure buying online and one of the main causes is the impossibility of trying the pieces and evaluating its fit (Apeagyei and Otieno 2007). The consumers enjoy the comfort that Internet shopping provides, but there will always remain some doubts about the appropriate fit of the pieces. Fan et al. (2004), refer some factors related with problems associated to the fit and performance of the pieces, like the gaps in standardization; problems with the sizes and rules of grading; faults in the development of patterns; conflicts of manufacture and the perception of the body by the consumer itself. To surpass this obstacle some companies have incorporate 3D simulation technology on their website, where the consumer can visualize a virtual model, and be able to choose with a greater level of confidence. This enhances the consumer or buyer to be near to the finished product and its real appearance, implying a decrease in devolutions and in an increasing of the level of satisfaction. The use of these systems raises the importance of the virtual try-out and the customer relationship management, since it provides the companies with an opportunity to increase sales as in physical stores (Apeagyei and Otieno 2007) and (Speer, 2002). In a market where the consumer can’t feel the touch of the pieces and its comfort, a more accurate description of fabrics provided by this type of applications with specific report and real images could lead the consumer to correlate with past tactile experiences. Nowadays, a pioneering company having singular products with the possibility to compete in price, quality, flexibility and delivery, can enter in the global market using the Internet (Kotler et al., 1999; Kumar 2004). With a small investment and a limited amount time to publish a web site on the Internet, a company can break barriers of entrance in the market with just a fraction of the resources needed to build an off-line store. However, to efficiently compete on the Internet, companies need to invest in the main areas previously referred and also guarantee the high quality services to the consumer (Kumar, 2004). 114 IADIS International Conference e-Commerce 2009 1.3 The Platform Interface for Customized e-Commerce Having developed an interface to virtual markets supported on information technology a company ends up having more freedom in space organization and design, namely in products presentation and the tools used to promote the interaction and choice. All alternative decisions must be accounted for, as they will have a direct influence in the way consumers will interact with the website. Space distribution, structure, colours, usability and accessibility will be reflected in the consumers’ response. For Bellman et al. (2006) the response will be positive if users can easily search and find information about the product they desire, and find no special difficulties to make a choice from the extensive range of services associated (evaluation of prices, independent opinions, more information about the product). In this process of developing an interface for the mass customization e-commerce platform the company can also involve the consumer in other levels, which goes beyond the mix of parts to make clothes, specifically sharing ideas, proposing new designs, materials or even ways of displaying the product. This process is not difficult, if the consumer feels that his cooperation can provide some added value and some prize could be earned. Some websites give money prizes and copyrights to the best designs and innovations, to the most voted proposals (Piller, 2008). 2. METHOD 2.1 Research Design and Data Collection Having as main purpose the development of the theoretical bases for the definition and implementation of an online e-commerce platform for customized clothing products, a study was conducted in order to analyze consumers’ behaviour in the electronic commerce of clothing, and their attitude regarding the design and fashionable trends through an online survey. The survey was made to gather information on three main topics: the purchase behaviour, e-commerce adoption/use and fashion trends (Ribeiro, 2008). This paper will exclusively address the consumer’s behaviour and e-commerce components, as they are the crucial ones for our current objective. The decision to develop this survey was due to the need to identify and define a profile of the proper market segments that could be successfully targeted by a customized clothing sales platform on the Internet. The survey focused online consumers, i.e., those who have already made purchases in online stores, and those who intended to do it. The initial questions were formulated in order to select and include only consumers that match the desired profile. Two groups of questions were included, one related to purchase behaviour and another to electronic commerce. The majority of the questions were multiple choice questions with one open answer option. The survey was created and distributed in digital format, through e-mail distribution and Google docs form utility. For e-mail distribution a written cover letter was prepared to stimulate the filling and diffusion Word file attached. Data collection began on May, 7, 2008 and ended on May, 31, 2008. After collection, data was coded and analyzed with SPSS 16 statistical package. 2.2 Results and Discussion From the initial sample of 366 answers, only 197 were retained for analysis, specifically those who have already made purchases of clothes over the Internet and those intended to do it in the future. As can be seen 46.17% of the respondents do not perform online purchases of clothes. From the 197 subjects that have already performed purchases and those that have never performed but intend to buy, 34.4% where between 24 and 28 years, followed by the group of 18 to 23 years with 23.4%, and from 29 to 33 years with 18.2%, being the remaining distributed among the other intervals. This result is in consonance with the National Inquiry of Consumption of the Families of the Acxiom (2007), where it is revealed that 40 percent of the entire number of purchases on the Internet was made by the age group of 18 to 35 years. 115 ISBN: 978-972-8924-89-8 © 2009 IADIS The majority of respondents were female (73%) and students (34.7%). Looking at 2007 conjoint report from INE and UMIC 2007, the majority of Internet users are high school students or hold a university degree. The Cetelem (2008) study shows that most of the Internet users maintain administrative, scientific, and intellectual professions, a tendency that was confirmed in the present investigation. The evaluation of consumers’ shopping habits reveal that 51% buy clothes at least every month, but an expressive majority (94% - 184 respondents ) indicate that they prefer to buy in traditional brick-and-mortar shops and just 4 respondents say that prefer to do it over e-commerce sites. The main reasons to visit clothing websites are presented in Table 1. Table 1. Reasons for Visit Brand Clothing Websites Reasons Check up the news Analyze the trends Look out for information about pieces of clothing Buy on-line Find contacts and store addresses Responses (%) 41 23 20 10 6 Consumers that had visited sites of clothes (54%), indicate that the main reason of the visit was to check up news and trends, and only 10% refer the specific intention of buying. This result clearly points to the advantage that could arise from the improvement of e-commerce platforms make them more user-friendly and promote the buying. The search for information seems to be the main motivation to use the Internet (see Table 2). This results stresses the importance of having a good online description with detailed information about the products, its characteristics and a good presentation, a fact that has been mentioned in studies of consumer‘s behaviour in virtual markets (Li et al., 2000). Table 2. Consumers and Buying Clothing in Internet Answers Uses Internet to find more information about the characteristics of the pieces to buy Uses Internet to find stores where he can find the pieces wanted Uses Internet to shop on-line Incidence (%) 43 31 26 In our study, the majority of the consumers do not perform purchases of clothing in electronic markets (69%), but a part of those (38%) is willing to start to buy if a set of conditions is present. This study reflect the general results of Portuguese consumer’s behaviour (INE and UMIC, 2007; Anacom, 2008; Cetelem, 2008) pointing that e-commerce still has a great potential to grow. Women seem to be more confident in buying clothes from e-commerce platforms, which is good for the development of mass customization business on Internet (see Figure 1). This can be justified by the relationship that females have with clothes and fashion in general. Women understand the fashion language more easily and show a personal relationship with clothes, which leads them to be more interested in a mass customization e-commerce interactive platform. Figure 1. e-Commerce Shoppers Versus Gender (%) We found that the main factors that encouraged consumers to use clothes e-commerce websites were: the possibility of choice without time waste with others products out of interest, the level of information 116 IADIS International Conference e-Commerce 2009 provided by websites compared with traditional printed catalogues, and the possibility to find pieces that are not available in traditional shops. The main reasons identified in our study are concordant with the tendencies of consumption of mass customized products stated by several authors (Pine and Gilmore, 1999; Anderson et al., 2002; Piller, 2004). Consumer feels somehow dissatisfied with the current offerings but they have a clear perception of their desires. On the other hand we can’t forget that there are still obstacles to complete adoption of fashion and clothes e-commerce that are mainly related with the need of a sensory experience in the purchase of clothing, i.e., the touch, the need to feel and determine performance of the clothes (Li, Daughrty, Biocca 2001). A list of prominent parameters and conditions that would promote e-commerce acceptance for fashion industry was compiled (see Table 3). These results suggest that the industry should pay attention to a set of parameters since the answers were almost equally distributed by the several factors. This implies that all of these parameters are important to attract consumers. However, some parameters stand out from the set, those related with return policies and process, and the existence of explicit and real images of the product to sell, namely the aspect of the fabrics and their performance. These answers show the lack of confidence of the consumer regarding the trueness of the pieces presented on e-commerce platforms. So it’s crucial to ensure the guarantees, namely for the return policies and the veracity of the products presented, in order to transmit security of the purchase experience in all aspects, obtaining higher levels of satisfaction (Fan et al. 2004). Table 3. Conditions that Would Promote the Acceptance of e-Commerce Conditions Easy return process and without complication Real images, showing fabric details and performance Different clothing from the stores Quality guarantee True made to measure clothing (on-line measures in a virtual model) Security payment The website have partnerships with official entities that can guarantee the existence of the business Let the consumer make part of design process by choosing colors, functionality and others True information Post-sale service On-line service to control the selling service Total Incidence 93 86 78 76 75 74 64 60 60 51 46 763 % 12,2 11,3 10,2 10,0 9,8 9,7 8,4 8,0 8,0 6,7 6,0 100 3. THE E-COMMERCE PLATFORM GUIDELINES As it has been referred in this study the electronic commerce is growing and will, in the future, hold an important position in the commercialization of all kind of products and services. Thus, it is important to think about taking part on this distribution channel and develop ways of attracting, retaining and satisfying consumers’ needs and desires, which can be accomplished with a structured construction of e-commerce platforms and its adequacy to product or service characteristics. With the survey, we get together a set of information’s with the purpose of developing a proposal for an online commerce platform of clothing and fashionable accessories based on mass customization with the necessary characteristics to succeed. It was possible to identify some key points to attract consumers to online market and e-commerce platforms. The main points revealed by this study, about the desired characteristics to include in an e-commerce platform of customized clothing were structured in four groups: 1) General Characteristics; 2) Customization; 3) Products; 4) Purchase experience. 3.1 General Characteristics General characteristics include navigation and client services issues. The navigation is one of the most important aspects in an e-commerce websites, for clothes or any other product. The consumer has to understand how the whole process works, starting with the search and choice of the pieces and ending in the purchase (Bae, May-Plumlee, 2005). From the analysis of several platforms of electronic commerce (Ribeiro, 2008), some factors related to navigation were identified, namely what should exist in a website to 117 ISBN: 978-972-8924-89-8 © 2009 IADIS assure its success, such as: short loading times; Information and comprehensible steps to follow; Absence of errors and incompatibilities with different browsers; to be accessible to all kinds of persons. Some services are essential to guarantee the success of the buying process and the maximum satisfaction of the client, such as: the change of idiom; loyalty - client account or loyalty card; a detailed guide explaining how to take measures in order to achieve the best fit, whether manually or with the help of 3D scanners; an assistant to guide and help in the first visit; a search option; a shopping basket; sales support and after-sales service; several payment options and security certificates; clear return policies and procedures; order tracking; clothes maintenance tips; the possibility of require piece transformation after purchase; trends shows; the possibility of see and buy models built by others consumers; voting for the designs; newsletter; blog; suggestions related with fashion and cultural events and finally on-site music. 3.2 Mass Customization Customization is about small prices and pieces that suit to the consumers’ desires of individuality, modularity, co-design and body fit (Prahalad and Ramaswamy, 2004; Piller 2004). The consumer wants more and more unique products, personalized to his needs. By this reason, investing in the development of an ecommerce platform of mass customization makes all the sense. The number of consumers that visit clothing websites use them especially to find information about existing pieces in brick-and-mortar shops, and consequently the sales volume of those websites tends to be low, which justifies that if a company starts operating in the electronic market it should offer customized products in order to avoid competing with their own physical stores. Internet, as distribution channel fits perfectly to this new strategy of production that requires the interaction between technology, products, consumers and processes. This idea is therefore reinforced by the answers to this study where consumers justify the use of this type of shops by the possibility of choosing only the products they desire and those they can’t find in traditional shops. 3.2.1 Customization Integrated in an On-line Website In existing websites where the consumer has the ability to build his own pieces and personalize them (ex: http://www.styleshake.com/, see Figure 2), the process of customization follows several steps since the creation of a database with relevant information’s (measures), piece definition, the part list to build it, and afterwards the choice of colours, woven, and decorative details. Some elements related with the product are key points to the success of implementation of mass customization websites, such as modularity, co-design and the fit to consumer’s body. 3.2.2 Modularity Those are products whose output process has modules as a base, in this case adaptable moulds to different types of pieces and ensure the multi functionality and the combination of different items to change the aspect of the piece, such as transform pieces according to the season. Modularity is also a fundamental element to give the consumer the opportunity to build clothes mixing different modules. To the company, modularity permits the costs control of the final piece, therefore those modules are normally mass produced, just being necessary its adaptation. 3.2.3 Co-design With co-design consumer can integrate the development of its pieces at early stages, either in the design or in the choice of modules already developed. In our sample almost the totality of respondents considers that an exclusive and unique design increases value of clothing and when questioned about the opportunity of having an active role in the creation the majority answered affirmatively (Ribeiro, 2008). In a more advanced level of co-design, the consumer shares ideas and new designs with the company, helping with the creation development of pieces to his taste and desires , but that can also satisfy others clients. 3.2.4 The Body of Consumer With 3D scanners technology is possible to provide this service which enables the manufacture of pieces entirely customized, that fit perfectly and as a result turning the purchase process more realistic and viable. This is crucial in this process where the distance between the consumer and the company decreases and 118 IADIS International Conference e-Commerce 2009 where the company is committed to offer customized and personalized products, to address issues of how the pieces will fit the consumer’s body. Figure 2. Style Shake (http://www.styleshake.com/) 3.3 Items to Include in the Customization Process The pieces chosen as possible options for implementation in an e-commerce platform for customized products comes from the questionnaire carried out and from several other sources such as design projects, and website analysis (Ribeiro, 2008) (see Figure 3). Product Profile Public Style Type of pieces Fabrics Colours Decorative elements Feminine Casual and urban Accessories, jackets dresses, tops, shirts, trousers, skirts) – 2 to 7 options 5 to 8 options 10 to 15 option Applications and printings – 10 to 15 options Separated parts that can be mixed together and be as 2 or more pieces of clothing - 20 to 50 options Figure 3. Characteristics of Products One important aspect in an electronic commerce platform is the quantity of products available. As Wind, Rangaswamy (1999) and Haugtvedt et al. (2005) referred, the bigger the quantity of products and options of choice the bigger the consumer becomes confused and frustrated, making it impossible for him to finish the buying process. Consequently the number of options were chosen according to the analysis made to customized online platforms, and are expected to be a reasonable number of choices that permits a wide range of combinations and pieces diversity but simultaneously don’t encourage the consumer to give up and leave. 119 ISBN: 978-972-8924-89-8 © 2009 IADIS 3.3.1 Clothing The type of clothing to sell on this kind of platform would have to target the female segment, because in the survey carried out we found a bigger enthusiasm from this segment in accordance with previous studies (Bakewell et al., 2006). The urban, alternative clothing, nightwear and casual appear as the ones that would be bought more easily. According to our survey these are the categories of clothing identified by the majority of consumers. Consequently, urban and casual clothing are, in our opinion, the right choice to be sold on an e-commerce platform for customized products, which is reinforced by the observation made of several websites. The pieces developed should follow principles of modularity, which necessarily incurs in multifunctional options. According to the consumers inquired, the most desired multifunctional options are the existence of separated parts that can be conjugated (ex: change of sleeves) and fulfil the function of two or more items of clothing (ex: dressed and skirt; trousers and bag). The accessories and the coats are the pieces consumers prefer to be multifunctional, namely by its capacity of transformation and inherent adaptation to diverse situations and seasons. 3.3.2 Fabrics, Colours and Decorative Elements Denim and other similar structures in cotton are the fabrics chosen especially by the knowledge of performance, the price, and by the type of platform and target defined. The composition of these fabrics will vary between cotton, biological cotton, wool, linen, bamboo, hemp and recycled polyester, in an average of 5 to 8 fabrics, varying with the nature of piece to produce. The colours go from brown to beige, but being available other neutral colours following the fashion tendencies. An average of 10 to 15 colour options will be available, that like the fabrics it can vary according to the nature of fabric. The consumers also show interest and desire to add extra elements, as sequins, decorative buttons, and embroideries, with the objective of personalize and differentiate an item of clothing, increasing its personal value. A set of elements are proposed, varying between 10 and 15 options, unfolding afterwards in diverse colours. 3.4 Buying Experience In the purchase experience field it is essential to try to match the virtual experience with the real one. Because of that the presentation of the products, should have real images with possibility of several levels of zoom and 3D rotation, and 3D simulation environments with virtual models and videos. The process should be straightforward, user friendly and fast. 3.4.1 Equalize Real Experience During the study one of the most important points was the consumer’s behaviour on virtual purchase environments. In the survey the majority of the consumers stated they preference for buying in real brickand-mortar shops and revealed a small preference for electronic environments, for perform clothing purchases. This could be justified mainly by the need to have direct contact with the product, try it and evaluate its performance (Apeagyei and Otileno, 2007). It is crucial try to match the sensorial experience of clothing purchase (Li, Daughrty and Biocca 2001), to give the feeling of pleasure and satisfaction that can occur in traditional shopping experience (Fan et al., 2004), but also give the consumer the security in all of the aspects of the process. With current technological evolution and level it is already possible to surpass these obstacles and turn the purchase experience more enjoyable. 3.4.2 Products Presentation The majority of the sites present the products on a traditional way in individual windows, with flat photographs of the pieces or on a mannequin. Zoom functionality is generally associated to the images, but rarely rotation and lively viewing. Presentation deficiencies are perhaps one of the main causes for the weak number of on-line purchases, because they aren’t attractive enough to persuade the consumer. Therefore, the existence of real images (photographs) and other forms of interactivity as the 3D rotation and by 3D simulation environments, virtual models, with the possibility of introduction of measures and physical 120 IADIS International Conference e-Commerce 2009 characteristics of the consumer are crucial to the success of the platform (Fan et al. 2004). The existence of small videos that allow the consumer to see the pieces in movement is also an important element to help perceiving the fit of the piece. 3.4.3 Products Functionalities Other functionalities associated to the presentation of the product, should be evident. Among those functionalities we found: the availability of information about the piece; colour and/or fabric options; the number of options to use simultaneously; size; price; the possibility to check stocks; suggestions of pieces to mix; the possibility to have recommendations; and to write an opinion about the product. With this description we intend to identify the main elements that should be applied on an on-line commerce platform of customized clothing in order to enhance its success. The choice and availability of all or just a part of the functionalities depends on the company resources and capabilities, trying to provide the best purchase experience with quality and security, but still profitable. 4. CONCLUSION With this analysis of consumer’s behaviour towards customized products purchase we can conclude that consumers are interested and give importance to elements associated with clothing customization: They desire individuality and unique products more congruent with their taste, personality and auto-image. Among those elements the multi functionality, the co-design and the fit to the body stands out. Besides that, price appears to be the main factor influencing the purchase decision. Customization as an integration of traditional mass production objectives with personalization potentiates the development of cheaper individualized fashion products is a good bet to achieve customers’ satisfaction and companies’ success. The majority of clothing e-commerce websites have an incomplete service, that can’t enchant consumers in the right way. This study identifies some areas and elements that are critical to mass customization ecommerce websites. The presentation of the product and the need to apply news technologies, such as 3D scanners and the virtual simulation are essential to guarantee a good consumer purchase experience, the nearest to the real one in traditional stores. This study and content proposal to an e-commerce platform of customized products suggests key points in diverse areas where more profound investigation and work for the implementation of this type of platform is going to be necessary, because, consumers still show some fear toward electronic markets. It is evident that in order to implement a project of this nature it is imperative to proceed carefully and with a deeper and detailed evaluation of market conditions and expected profits. Future investigations must be carried out to extend the knowledge on consumer’s behaviour for customized products in electronic markets trying to improve the relationship on between the company and the consumer; the development of alternative production systems that integrate mass production with customization and finally, contributions are necessary to wipe off some cost problems associated to the essential software and hardware. REFERENCES Anderson-Connell, Lenda Jo et al, 2002. A consumer-driven model for mass customization in the apparel market. Journal of Fashion Marketing and Management, Vol. 6, No. 3. Apeagyei, R. and Otieno, Rose, 2007. Usability of pattern customizing technology in the achievement and testing of fit for mass customization. Journal of Fashion Marketing and Management, Vol. 11, No. 3. Autoridade nacional de comunicações (Anacom), 2008. Informação estatística do serviço de acesso à Internet, 1º trimestre de 2008. Bae, JiHyun and May-plumlee, Tracci, 2005. Customer focused textile and apparel manufacturing systems toward an effective e-commerce model. Journal of textile and apparel, technology and management, Volume 4, issue 4. Bakewell, Cathy et al, 2006. UK generation Y male fashion consciousness. Journal of Fashion Marketing and Management, Vol. 10, No. 2. 121 ISBN: 978-972-8924-89-8 © 2009 IADIS Bellman, Steven et al, 2006. Designing marketplaces of the artificial with consumer mind: four approaches to understanding consumer behaviour in electronic environments. Journal of Interactive Marketing, Vol. 20. Cetelem, 2008. Dossier de Imprensa – O observador cetelem. Fan, Jintu et al, 2004. Clothing appearance and fit: science and technology. The Textile Institute, Woodhead Publishing Limited. Haugtvedt, Curtis P. et al, 2005. Online consumer psychology: understanding and influencing consumer behaviour in the virtual world. Lawrence Erlbaum Associates Publishers, London. Instituto nacional de estatística (INE), Agência para a sociedade do conhecimento (UMIC), 2007. A sociedade da informação em 2007. Kotler, Philip et al, 1999. Principles of Marketing, Second European Edition. Prentice Hall Inc. Kumar, Ashor, 2004. Mass customization: metrics and modularity. The International Journal of Flexible Manufacturing Systems, Vol. 16. Li, Hairon et al, 2001. Characteristics of virtual experience in electronic commerce: a protocol. Journal of Interactive Marketing, Vol..15. Lyengar, S. and Lepper, R., 2000. When choice is demotivating: can one desire too much of a good thing? Journal of Personality and Social Psychology, Vol. 79(6). Ma, Min-Yuan et al, 2007. A design decision-making support model for customized product colour combination. Computers in Industry, Vol. 58. Piller, Frank, 2004. Mass customization: reflections on the state of the concept. The International Journal of Flexible Manufacturing Systems, Vol. 16. Piller, Frank, 2008. Interactive value creation with users and customers. Pine ii, Joseph and Gilmore, H, 1999. The experience economy. Harvard Business School Press, Boston, Massachusetts. Prahalad, C. K. and Ramaswamy, 2004. Co-creation experiences: the next practice in value creation. Journal of Interactive Marketing, Vol 18. Ribeiro, Liliana, 2008. O co-design e a “mass customization” no desenvolvimento de uma plataforma on-line de comercialização de vestuário. Fashion Design Master Thesis, University of Beira Interior, Covilhã, Portugal. Wind, Jerry and Rangaswamy, Arvind, 1999. Customerization: the second revolution in mass customization. eBusiness research centre working paper. 122 IADIS International Conference e-Commerce 2009 TIME AND SPACE CONTEXTUAL INFORMATION IMPROVES CLICK QUALITY ESTIMATION Mehmed Kantardzic, Brent Wenerstrom, Chamila Walgampaya University of Louisville CECS Department J.B. Speed Building - Room 123 University of Louisville Louisville, KY 40292 Oleksandr Lozitskiy Microsoft Corporation 1020 102nd Ave NE Bellevue, WA 98004 Sean Higgins, Darren King Hosting.com 462 South Fourth Street Louisville, KY 40202 ABSTRACT Click fraud is a type of internet crime that occurs in pay per click online advertising when a person, automated script, or computer program imitates a legitimate user clicking on an ad, for the purpose of generating a charge per click without having actual interest in the ad’s content. We propose Collaborative Click Fraud Detection and Prevention system (CCFDP) V1.0, which integrates client and server side data to score the quality of incoming clicks. In this paper we detail the outlier detection module which describes clicks in terms of space and time context. This module compares past data with the current context as a preprocessing step. Our system will then combine the additional time and space characteristics with the characteristics of a click to score the quality of incoming clicks. We believe that no other commercial or research system for click fraud detection analyze comprehensively time and space context of each click for better estimation of click traffic quality. Some commercial solutions give only partial solutions expressed through their rules and triggers. We found 34.6% of clicks in an application to real data had outlying attribute-values in time and space. KEYWORDS Click Fraud, Click Fraud Prevention, Click Traffic Quality, Web Analytics. 1. INTRODUCTION A Web search is a fundamental technology for navigating the Internet and it provides information access to millions of users per day. Internet search engine companies, such as Google, Yahoo and MSN, have revolutionized not only the use of the Internet by individuals but also the way businesses advertise to consumers (Wang, Lee et al. 1998; Immorlica, Jain et al. 2005; Mahdian 2006). Typical search engine queries are short and reveal a great deal of information about user preferences. This gives search engine companies a unique opportunity to display highly targeted ads to their users. Many services such as Google, Yahoo and MSN generate advertising revenue by charging advertisers per click on advertisements shown. This business model is known as the pay-per-click model. In the pay-per-click business model, content providers are paid by the traffic they drive to a company’s website through online ads. There is an incentive for dishonest service providers to inflate the number of clicks their sites generate. In addition, dishonest advertisers tend to simulate clicks on the advertisements of their competitors to deplete their advertising budgets (Metwally, Agrawal et al. 2005). This fraudulent 123 ISBN: 978-972-8924-89-8 © 2009 IADIS behavior results in bad reputations and often extra costs. Generation of such invalid clicks either by humans or software with the intension to make money or deplete competitor’s budget is known as click fraud. There is no globally accepted mechanism to detect click fraud. Therefore, most of the search marketing industries optimize traffic using their own binary paradigms, dividing clicks into two categories, valid and invalid clicks, according to their own metrics (Clicklab LLC). These solutions are still not mature. The bottom line is, with the exception of a small percentage of obviously genuine or robot clicks, the vast majority of clicks simply cannot be classified to be either valid or invalid. Clicks could more accurately be scored by quality depending on available information. Neither commercial solutions for click fraud detection (usually based only on client side data) nor search engine solutions (mainly using server data) offer a complete solution (Ge, Kantardzic et al. 2005; Ge, Kantardzic et al. 2005). We initiated our work on the CCFDP system with an assumption that more data about each click collected from different sources will result in a better estimation of the click quality. The preprocessing phase of this system involves two main modules: a rule-based module and an outlier detection module. The rule-based module uses both a set of handcrafted rules and a database of bad data sources to produce a set of characteristic scores for an incoming clicks. The outlier detection module compares current clicks with a much larger aggregation of clicks to determine deviations from normal activity. Deviations are given a score by the outlier detection module. These scores are added to the click record to describe how a click may be suspicious in their time and space context. Dynamic advertising profiles use the aggregated scores from traffic sources to block future advertisements to particularly low quality sources of traffic. Countless dollars are saved by the advertiser through eliminating the lowest quality sources of traffic. We expect that clicks made during click fraud attacks will display detectable patterns (Daswani and Stoppelman 2007). Our outlier detection methodology uses attribute-value pair level aggregations so that we may detect repeats in attribute-value pairs such as user-agents, IPs, referrers and more. We maintain counts of all attribute-value pairs through the history of advertising campaigns to determine “normal” activity for each attribute-value pair. We then compare the aggregations of previous data with an aggregation of a current window of context. Our approach differs from most outlier detection methods that instead of outlier detection being the focus, we use outlier detection to extend the description of each click with time and space context. Using our outlier detection module we were able to score real-world click data for Hosting.com. Experimental data show a promising separation of high-quality and low-quality clicks. This paper is organized as follows: In section 2 an introduction to the pay-per-click model and click fraud categories are given. In section 3 a detailed discussion about the creation of baselines is given followed by our scoring mechanism in section 4. Section 5 discusses available solutions to detect click fraud in advertising. Section 6 discusses data analysis and results. Conclusions and future work are given in section 7. 2. PAY-PER-CLICK MODEL In a typical Internet traffic model, a Web user will first request a web page on the publisher’s site. The requested page is loaded along with the advertisement on the Web user’s browser. If the Web user clicks on an advertisement hypertext link (for example a banner ad or logo) on that page, the publisher redirects the Web user request to the commissioner’s server. The commissioner is an independent entity that has agreements with both the advertiser and the publisher and could be a search engine or other advertising agent. The commissioner logs the click for accounting purposes. The commissioner then redirects the Web user’s browser to the advertiser’s site. The publishers are paid based on the click traffic they drive to the advertiser’s web site. The commissioner earns a percentage of this revenue. Sometimes these payments are based on number of sales generated in the advertiser’s website, rather than the volume of traffic drives by the publisher (Ge 2008). Click fraud may occur in one of three ways (Kantardzic, Walgampaya et al. 2008). First, it could come from one or more human users. For example in developing countries a number of individuals may be hired for a nominal rate to simply click and reclick advertisements. Second, clicks may come from a robot actively perpetrating fraud on one or more advertisements. Thirdly, fraud may occur as a background process. For example, a user may inadvertently download some program which will cause clicks to be registered from that machine, without the users consent or knowledge (Daswani and Stoppelman 2007). 124 IADIS International Conference e-Commerce 2009 3. DYNAMIC BASELINES A number of definitions have been given to the term “outlier” depending upon the task at hand. For example Hawkins (Hawkins 1980) defines outliers by the following: “an outlier is an observation that deviates so much from other observations as to arouse suspicions that it was generated by a different mechanism.” This description points out two important points for our task. First, that an outlier should be considered suspicious. The deviation from “normal” data gives rise to suspicion, suspicion that some unusual mechanism has generated this data point. Second, what is considered normal must be modeled in such a way that points that deviate from what is considered normal can be detected and separated from the rest of the data. This section will describe how we model normal data, or in our case normal click traffic. Each click record is made up of a number of attributes such as browser, operating system, IP, referrer, etc. We model each attribute separately. We will determine normal behavior for referrer separate from all other attributes and browser separate from other attributes and so on. For each attribute we create histograms which maintain counts for the most common values in each attribute. These histograms are called baselines. An example of such a histogram can be seen in Figure 1 (left). In Figure 1 (left) one can see the number of clicks for each of four values of the referrer attribute. These histograms are then used to calculate the percentages for each value. In addition to absolute counts being maintained in the baseline, we also measure variance. Over time certain values were found to vary greatly in the percentage of traffic with a given value. For example traffic with Google as the referrer varied through out time from nearly 100% to less than 50% of traffic in a given day. This variance made comparisons with the baseline difficult. Variance is calculated using the formula to follow. This formulation of variance takes into consideration time periods varying in number of clicks. In the above formula each time period within the baseline is given by i. The xi is the count within that time period for the attribute-value pair. The ci is the total number of clicks within the time period. The value μ is the percentage of clicks for the current attribute-value pair throughout the entire baseline. Lastly, pi gives the percentage of clicks in the baseline from the current time period. Our baselines are used to identify outliers in incoming traffic among the attributes. Since we are identifying outliers based on context, we must necessarily accumulate clicks for a comparison of short term context with past context. These short term accumulations of clicks are called aggregated windows. An aggregated window is also treated as a histogram. An example of the comparison of an aggregated window to a baseline can be seen in Figure 1 (left and center). Figure 1. (Left) Global Baseline for Four Referrers as of 1/22/08. (Center) Aggregated Window for 1/22/08. (Right) Counts and Thresholds for Referrer on 1/22/08. Thresholds are Dark Gray and Counts in Light Gray We assume that percentages for a given attribute-value pair are distributed normally through time. An outlier is detected when a count for an attribute-value pair in an aggregated window is found in the upper 5% tail of the normal distribution, or is 1.645 standard deviations above the mean. We calculate the threshold in terms of number of clicks using the following formula to ensure whole number thresholds: 125 ISBN: 978-972-8924-89-8 © 2009 IADIS where is the percentage of the baseline made up by the attribute-value pair and σ is the standard deviation of that percentage in the baseline. Additionally, windowsize refers to the number of clicks in the current aggregated window. If the number of clicks in the aggregated window for a particular attribute-value pair is greater than the threshold, then it is considered an outlier. An example of calculated thresholds for can be seen in Figure 1 (right). Outliers are detected for each attribute in a given aggregated window. The results are then applied to the following context fields for each record: referrer, browser, operating system, country, ISP, and IP. If a record has an outlying referrer for the current aggregated window, it is reflected in the referrer context field. An example of an extended record can be seen in Table 1. Baselines are then recalculated adding in the current aggregated window to keep baselines up to date and for comparison to future aggregated windows. Table 1. Example Record after Outlier Detection Preprocessing. Record now Contains Server Side Data, Client Side Data and Context Based on a Number of Attributes Click Data Server Side Client Side 2 3 4 0 1 1 2 1 Referrer 0.18 Context Fields Browser OS Country 0.12 0.03 0.13 ISP 0.02 IP 0 4. CONTEXT-BASED ATTRIBUTE SCORING The CCFDP system scores each incoming click. The score will be a value in the range [0,1], where a zero represents no evidence of fraudulent behavior, and a one represents 100% confidence in the click being fraudulent. Within the CCFDP system, the outlier detection module provides relevant context to each attribute for scoring each click. Each click is provided a partial score for each attribute based on the variation of that click within the current context from normal behavior. Zero signifies that no evidence was found of suspicious activity for the given attribute-value pair in the current context. We will discuss two approaches for scoring an attribute-value pair when the count exceeds the calculated threshold. The first approach to scoring attaches a context score to a click record which conveys the extent to which a count has exceeded its threshold or the degree of suspicion for that particular count. Take for example referrer X. In the case where referrer X’s calculated threshold is 2, it would seem logical to give a higher partial context score if the actual count were 38 compared to 3. A different partial score is given to counts that far exceed their threshold, compared to counts that barely exceed their threshold. This provides more information to the overall scoring algorithm. This could allow for more refined overall scoring by providing a range of values for suspicious activity. This approach to scoring uses the difference between the attribute-value pair count in an aggregated window and the corresponding threshold. Then take that as a percentage of the total number of clicks in the aggregated window. A score in this first approach is given by the following formula: This type of scoring will be referred to as “variable scoring” by later sections. The closer to one the score gets, the more certainty given by the outlier detection system that something suspicious is happening. When the score is close to zero, then little evidence for suspicion is available. The second approach is to simply give a constant score to all attribute-value pairs exceeding their thresholds. Click fraud is perpetrated in a large number of ways, some approaches (Daswani and Stoppelman 2007) expect that a large number of clicks over a short period of time raises suspicions, and attempt a “lownoise click fraud attack”. 5. RELATED WORKS Several commercial solutions (Clicklab LLC ; Web Traffic Intelligence Inc. ; Tuzhilin 2006) and academic solutions (Immorlica, Jain et al. 2005; Metwally, Agrawal et al. 2007) exist for the problem of click fraud. None of the commercial solutions combine data from both the client and server side. Generally, the academic 126 IADIS International Conference e-Commerce 2009 solutions are incomplete and generally focus on one specific fraudulent attack. We propose a new system for grading the quality of clicks incorporating both client and server side data. Several approaches to the outlier detection problem have been proposed for a variety of purposes. Some of these approaches include statistical profiling using histograms (Javitz, Valdes et al. 1991), statistical modeling (Yamanishi, Takeuchi et al. 2000), and clustering (Eskin, Arnold et al. 2002). Some similar approaches to our baseline approach come from the intrusion detection research. For example Javits et al. (Javitz, Valdes et al. 1991) discuss a real-time intrusion detection system which produces a score per attribute, where attributes could be categorical or numerical. Rare categorical events were given high scores or were more likely to be considered outliers. Anderson et al. (Anderson, Frivold et al. 1995) explain at a high level a system combining both rules and anomaly detection for detecting network security violations. Denning (Denning 1987) describes a real-time intrusion detection system which detects large variations based on audit data. Endler (Endler 1998) detects intrusions using a histogram classifier trained on labeled data. Ali and Scarr (Ali and Scarr 2007) developed a robust model to detect outliers and robots based on their number of clicks from the home page. Qu et al. (Qu, Vetter et al.) devised a multi-step process for intrusion detection. First the rates are determined per event. Then the deviation of their scoring routine is calculated and finally a threshold value is determined for the entire process. Overall, these papers (Qu, Vetter et al. ; Denning 1987; Javitz, Valdes et al. 1991; Anderson, Frivold et al. 1995; Endler 1998; Yamanishi, Takeuchi et al. 2000; Eskin, Arnold et al. 2002; Ramadas, Ostermann et al. 2003) show that a number of approaches have been shown to be effective in determining outliers using frequency counts. These are stand alone outlier detection systems. Our outlier detection system is a data preprocessing step which enables an enhanced description of clicks. 6. EXPERIMENTAL RESULTS The data which will be used for the experiments to follow comes from the paid traffic of two very different websites. All of our experiments use click data from Hosting.com’s website. Hosting.com is a global company which provides hosting solutions to “business critical data assets” as explained on their website. We also created thebestmusicsites.org, a single webpage only displaying advertisements. It was created to attract fraudulent traffic. Currently we have data for Aug. 2007 to June 2008 for Hosting.com and Jan. 2007 to Aug. 2007 for thebestmusicsites.org. Traffic was filtered before being run through the outlier detection module. First, only paid traffic is being monitored. This includes ads run on search results and network partners of Google, Yahoo, MSN and Miva. Next, we removed all known robots from the traffic. A known robot is a robot that declares itself in the user agent field. Known robots are generally removed from paid traffic before an account is charged. Lastly, to prevent the distortion of baselines, repeat clicks from the same visitor (defined by IP and user agent together) during the current window were removed. Large numbers of repeat clicks by the same user would falsely inflate all attribute-value pairs causing false alarms and distorting normal behavior. After filtering Hosting.com’s click data, 9,252 clicks remained of the original 1,083,037 clicks. A majority of the original clicks recorded are not coming through paid sources. Experiments were conducted by taking the click data that we had and simulating the progression of days. For each available date in the data a baseline would be composed of all data previous to that date and the current date would be used as the current window in which outliers would be detected. Unique values per attribute per window is the level of granularity at which outliers are detected. Each click receives a score per attribute and the scores are combined to make up a partial score for the outlier detection module to later be combined with scores from other modules for each click. In these experiments values not found to be outliers are given a score of zero. The effectiveness of each experiment was based on how well traffic with known user activity was divided between the clicks with outlying values and clicks without outlying values. Clicks with user activity are more likely to be made by human users and therefore are more likely to be valid paid clicks. User activity is recorded through JavaScript functionality which recorded when users used the keyboard or moved their mouse. As an example, consider paid traffic from known robots compared to the rest of the paid traffic. Only 0.7% of traffic generated by known robots had recorded user activity, while 24% of the rest of the paid traffic 127 ISBN: 978-972-8924-89-8 © 2009 IADIS had some form of user activity recorded. Thus we would expect traffic from known robots is of lower quality than traffic not from known robots. Table 2. Parameter List for the Outlier Detection Module Parameter Aggregated window length Global baseline history length Minimum global baseline rate Method of scoring Description What defines the start and end of the aggregated window? How long of a time period is encapsulated by the global baseline? What is the minimum allowable baseline rate to be stored in the database? How will each outlier be scored? The user defined parameters for the proposed algorithm can be seen in Table 2. For the experiments explained here we have set the “aggregated window length” to 24 hours. The global baseline history length will be set to a year. The “minimum global baseline rate” used is 0.5%. In our experiments to follow we will compare two proposed methods of context scoring. Lastly, we will test the module with the parameters tuned by previous testing on all of Hosting.com’s paid clicks to determine the usefulness of our approach. 6.1 Scoring Method for Outliers in Aggregated Windows When an outlier is found, there are a number of ways to determine what score to attach to that attribute-value pair. We have mentioned two scoring approaches in Section 4. The first was to use a constant for the score, and we will call this method “constant scoring”. The second approach was to determine the degree to which the count exceeded the threshold, will be referred to as “variable scoring”. We now add two additional rules to the variable scoring technique, (1) that if the window size was less than 20, that a value of 20 be used for the window size, and (2) that scores that exceed the score of 0.2 be reduced to 0.2. We found that when the window size is only 3 or 4 clicks, large scores could result merely due to the window size. Putting a maximum value on the score made very little difference in the outcome, but prevented the rare case of a single attribute-value pair receiving extreme values such as 0.6 which would have a large impact on an overall score. These rules serve as a saturation function to prevent any single attached partial score from having too much influence on the final score of a click. Figure 2. Comparison of Scoring Methods by User Activity Rates. Constant Scoring is on the Left and Variable Scoring is on the Right Table 3. Break Down of User Activity Percentages for Scoring Techniques by Scoring Bins (Another View of Figure 2) Table 4. Outlier Detection Performance Outlying Column Count referrer 1,714 ISP 924 browser 922 OS 834 country 504 IP 84 total 3,200 Constant scoring used the score of 0.2. The value 0.2 was only used for testing this system. Context Score 0 (0, 0.05] (0.05, 0.1] (0.1, 0.15] (0.15, 0.2] total 128 Constant Scoring 1625 / 6052 (26.9%) 591 / 3200 (18.5%) 2216 / 9252 (23.9%) Variable Scoring 1625 / 6052 (26.9%) 295 / 1319 (22.4%) 139 / 757 (18.4%) 44 / 318 (13.8%) 113 / 693 (16.3%) 2216 / 9252 (23.9%) Percentage 18.5% 10.0% 10.0% 9.0% 5.4% 0.9% 34.6% IADIS International Conference e-Commerce 2009 A comparison of the two scoring methods by user activity rates can be seen in Figure 2 and Table 3. Figure 2 depicts two mosaic plots showing user activity percentages according to context score. Constant scoring had two score levels 0 and 0.2. The variable scoring provided a large number of maximum scores. We split the scores above zero into four levels. As can be seen from the graph and table, as the score increases the percentage of clicks with user activity decreases, except for the range (0.1, 0.15]. This general trend of decreasing user activity rates by maximum context score shows that the variable scoring approach provides more useful information than constant scoring. 6.2 Overall Performance of the Outlier Detection Module In this section we show the results of the tuned outlier detection module on the Hosting.com data. The goal of this research was to create a preprocessing step which would add context in time and space to each new click. The previous experiment suggests that context scoring provides better separation of low and high quality clicks, where quality is judged by user activity rates. In Table 4 we show the percentage of each attribute found to be an outlier. Overall we see 34.6% of paid clicks on Hosting.com had an outlier in one or more attributes. The attribute with the most number of outliers is referrer. IP had the fewest number of outliers. This should be the case, because we find the most number of unique values in IP. This is why country and ISP are necessary to group IPs geographically. Figure 3. (Left) Timeline of Hosting.com Clicks from January 2008 through June 2008. (Right) Clicks with Outlying Attribute(s) from May 27 through June 20, Showing the Two most Prevalent Referrers during this Period To gain additional insight into our traffic source, we plot click counts over time in Figure 3 (left). Clicks which have one or more outlying attributes are labeled as “outliers”. From these two graphs we can see that a large number of outliers occurred during the end of May and the start of June. During this period there were a number of factors that would contribute to outlying scores. First, there was a major shift in keywords selected. Next there was a spike in traffic over a short period of time. This could reflect a natural burst in traffic or could be the result of click fraud. Upon further review 45% of the clicks with non-zero, partial context scores during late May, early June came from two, never before seen referrers, see Figure 3 (right). These two referrers were more common than any referrer except Google and an empty referrer, which is suspicious. More information is required to determine the level of suspicion. From these results we see that the outlier detection module is able to add new context-based information. This new information provides a better description of the clicks than is currently available through commercial and non-commercial means. This added information will aid the CCFDP system in better scoring the quality of clicks. 7. CONCLUSION In this paper we outline a methodology for preprocessing paid clicks for the extraction of additional contextbased information (in time and space). The improved description of incoming clicks will improve the analysis and estimation of click quality. We tested this module on a years worth of advertisement click data 129 ISBN: 978-972-8924-89-8 © 2009 IADIS for Hosting.com and found outliers in one or more attributes of 34.6% of clicks. These clicks show evidence of suspicious behavior, but will require the entire scoring process to determine the quality of the click. We are continually collecting data from different websites to test our CCFDP system and we plan on testing the entire integrated system with this new data. We plan on comparing our system to a number of commercial solutions. This will enable a more accurate understanding of the improvements that we are proposing in our CCFDP system. ACKNOWLEDGMENT This research has been partially funded by National Science Foundation (NSF) under grant #0637563 and Kentucky Science and Technology Corp. (KSTC) under grant #KSTC-144-401-07-018. REFERENCES Ali, K. and M. Scarr, 2007. Robust methodologies for modeling web click distributions. WWW, ACM Press New York, NY,USA. Anderson, D., T. Frivold, et al., 1995. Next-generation Intrusion Detection Expert System (NIDES): A Summary, SRI International, Computer Science Laboratory. Clicklab LLC. from http://www.ClickLab.com. Daswani, N. and M. Stoppelman, 2007. The anatomy of Clickbot. A. Proceedings of the first conference on First Workshop on Hot Topics in Understanding Botnets table of contents. Denning, D. E., 1987. An Intrusion-Detection Model. IEEE Transactions on Software Engineering 13(2): 222-232. Endler, D., 1998. Intrusion Detection Applying Machine Learning to Solaris Audit Data. Proceedings of the 14th Annual Computer Security Applications Conference, IEEE Computer Society. Eskin, E., A. Arnold, et al., 2002. A Geometric Framework for Unsupervised Anomaly Detection: Detecting Intrusions in Unlabeled Data. Applications of Data Mining in Computer Security, Kluwer. Ge, L., 2008. CCFDP V. 1.0: Final report for STTR-NSF project #0637563. Ge, L., M. Kantardzic, et al., 2005. CCFDP: Collaborative Click Fraud Detection and Prevention System. International Conference on Computer Application in Industry and Engineering. Honolulu. Ge, L., M. Kantardzic, et al., 2005. Collaborative Click Fraud Detection and Prevention System (CCFDP) Discovers Software-Based Click Fraud. IADIS e-Commerce. Porto, Portugal. Hawkins, D. M., 1980. Identification of outliers. London, England, Chapman and Hall. Immorlica, N., K. Jain, et al., 2005. Click Fraud Resistant Methods for Learning Click-Through Rates. Workshop on Internet and Network Economics (WINE), Hong Kong. Javitz, H. S., A. Valdes, et al., 1991. The SRI IDES statistical anomaly detector. Research in Security and Privacy, 1991. Proceedings., 1991 IEEE Computer Society Symposium on: 316-326. Kantardzic, M., C. Walgampaya, et al., 2008. Improving Click Fraud Detection by Real Time Data Fusion. IEEE International Symposium on Signal Processing and Information Technology, Sarajevo. Mahdian, M., 2006. Theoretical challenges in the design of advertisement auctions, The Capital Area Theory Symposia, University of Maryland. Metwally, A., D. Agrawal, et al., 2005. Duplicate Detection in Click Streams. Proceedings of the 14th WWW International World Wide Web Conference: 12-21. Metwally, A., D. Agrawal, et al., 2007. Detectives: detecting coalition hit inflation attacks in advertising networks streams. Proceedings of the 16th international conference on World Wide Web: 241-250. Qu, D., B. M. Vetter, et al., Statistical Anomaly Detection for Link-State Routing Protocols. contract 30602(96-C): 0325. Tuzhilin, A., 2006. The Lane’s Gifts v. Google Report. Wang, H., M. K. O. Lee, et al., 1998. Consumer privacy concerns about Internet marketing. Communications of the ACM 41(3): 63-70. Web Traffic Intelligence Inc. from www.clickalyzer.com. Yamanishi, K., J. Takeuchi, et al., 2000. On-Line Unsupervised Outlier Detection Using Finite Mixtures with Discounting Learning Algorithms. KDD. 130 IADIS International Conference e-Commerce 2009 THE BRAZILIAN MOBILE DIGITAL CONTENT MARKET: AN OVERVIEW Marcelo Cortimiglia, Filippo Renga and Andrea Rangone Politecnico di Milano - Department of Management, Economics and Industrial Engineering Via Giuseppe Colombo, 40 – 20133, Milan, Italy ABSTRACT Mobile telephony is one of the most important segments of the global telecommunications industry. As a result of subscription saturation in some markets and intense competition for market share in others, voice revenues stagnate and mobile telephony network operators (MNO) find themselves increasingly dependent on data revenues, such as those derived from the creation, distribution and use of digital content. This reality is especially true in emergent countries like Brazil, China, Russia and India, which are frequently referred to as the mobile battleground for the next decades. Furthermore, there is a lack of empirical academic studies about specific mobile data services, such as the mobile content segment, which could expand the understanding of subtleties in topics such as value networks (at firm level) and service adoption (at individual level). In this context, this paper presents the results of an empirical study to analyse the offering of mobile digital content in Brazil. A two-phase mixed methods exploratory research was conducted, consisting of a quantitative web and WAP-based survey of 3,678 mobile content services complemented by a deep case study on the Brazilian Mobile Content Value Network based on 39 semi-structured interviews with key market experts. KEYWORDS Mobile Commerce, Emerging Countries, Digital Content. 1. INTRODUCTION Telecommunications is one of the most important businesses today. Estimated 2008 worldwide revenues accounted for circa U$ 1.7 trillion, a number expected to grow to U$ 2.7 trillion by 2013 (The Insight Research Corporation, 2007). Of special interest is the mobile telephony segment. According to the GSM Association, there are currently more than 4 billion mobile connections worldwide, reaching 6 billion by 2013 (Mika, 2009). However, there is a global trend for stagnant or even decreasing voice revenues, leaving data revenues to fuel the growth of the industry. While this was already pointed out at the beginning of the 2000s by Maitland et al. (2006), it is an even more strong proposition today, as the mobile industry suffers in the wake of global economic recession (Richtel, 2009). A natural response is that of heightened interest in data business, an umbrella term that encompasses segments such as mobile marketing, mobile content, mobile payment, mobile search, and mobile services (Dunnewijk & Hultén, 2007, Kallio et al., 2006). Among these, mobile digital content is one of the most promising. In 2007, mobile music worldwide was worth U$ 9.3 billion, and mobile games over U$ 5 billion (Netsize, 2007). Holden (2008b) expects that the global mobile content market will be worth U$ 167 billion by 2013. Evidence of the same trends can be found in emerging countries, especially in the so-called BRIC (acronym for Brazil, Russia, India and China). The combined telecommunications market for these countries was worth U$ 256.1 billion in 2007, with more than 1 billion mobile subscribers, and voice average revenue per user (ARPU) will decrease from U$ 8.05 in 2008 to U$ 5.90 in 2013, while data ARPU will increase from U$ 1.94 to U$ 2.49 in the same period (Kitson, 2008). Particularly in Brazil, evidence for these trends can already be seen: network operators' gross data revenues for the 3rd quarter of 2008 accounted on average 9.6% of gross service revenues, up 2 p.p. year-over-year (Teleco, 2009d). Similarly, Frost & Sullivan (2008) highlights the importance of mobile content in Brazil by estimating at 37.7% the number of mobile 131 ISBN: 978-972-8924-89-8 © 2009 IADIS subscribers that used at least one mobile content service during 2007, while forecasting a 34.9% compound annual growth rate between 2007 and 2013. Mobile Content is also relevant from an academic point of view, as there are intriguing unresolved academic issues in the field of business strategy applied to the mobile digital content industry. In the context of a changing mobile phone landscape characterized by a large number of firms establishing increasingly complex relationships between them, there have been discussions about a product- or service-focused value network structure replacing the simpler firm-focused value chain structure (Peppard & Rylander, 2006; Funk, 2009). In this context, characteristics of value networks become highly dependent on the type of product or service being considered. This is a further motivation for studying in detail a particular type of mobile data service, such as the mobile content segment. Moreover, at the individual level of analysis, it seems that the topic of mobile services adoption is still lacking a substantial empirical base. In particular, Bouwman et al. (2008) highlighted the importance of taking into account the specific characteristics and user values involved with different types of mobile data services when investigating its adoption. This kind of problem is even more pressing when considering the impact of local economic, legislative, technological and cultural realities in mobile data industry segments (Funk, 2006, 2007; Lu et al., 2007; O'Donnel et al., 2007; Bouwman et al., 2008). Although it is clear the business and academic relevance of the mobile content industry, especially in emerging countries, up to now there are no systematic studies about the topic in such countries. In light of these considerations, the objective of this paper is to present a descriptive overview of the Brazilian mobile content business environment, with special focus on the mobile content offering. 2. LITERATURE OVERVIEW In this section, it is presented an overview of published research on the topics of mobile business and commerce and mobile digital content, as well as an overview of the Brazilian mobile telecommunications sector. 2.1 Mobile Business and Commerce In many sources, the expressions mobile business and mobile commerce are used quite interchangeably. However, some argue that mobile business is focused on the impact of mobility on organizational structures, strategies and processes (Scornavacca et al., 2006, Barnes & Scornavacca, 2007). It is, evidently, the realm of business to business (B2b) and business to employee (B2e) applications, outside the scope of this paper. In literature, mobile commerce is usually referred to as a subset – or an extension – of electronic commerce. A classical definition of mobile commerce was provided by Barnes (2002): “any transaction with a monetary value – either direct or indirect – that is conducted over a wireless telecommunication network”. It includes but does not limit mobile commerce to the purchase of goods or services, opening up the case for applications such as mobile payment and advertisement. It is still a wide definition, but much more rooted in B2C. There are, however, even narrower definitions, like the one by Maitland et al. (2006): “buying and selling of goods and services through wireless hand-held devices”. It is this narrower definition, particularly adequate to the purpose of researching mobile content, which will be adopted here. 2.2 Mobile Digital Content A practical concept for digital content is proposed by Loebbecke and Huyskens, 2007, who define it as data, information or knowledge products traded exclusively through online networks and characterized by indestructibility, transmutability, reproducibility, intangibility (understood specifically as immateriality, or absence of any tangible components), and by the fact that usually product quality can be learned only by actually using the good. This definition of digital content is valid independently of the technological platform for content distribution and use. Therefore, to define mobile digital content as a particular subset of digital content it is enough to limit its creation, distribution and use to that technology. In light of these considerations, the definition of mobile digital content adopted in this paper is that of Feijóo et al., 2009: “mobile content refers to the creation-production, distribution-access and consumptionuse-interaction of content, be it creative or processed information, on a mobile platform, consisting at least of a mobile device and/or a mobile network”. Consequently, person-to-person messaging services, payment 132 IADIS International Conference e-Commerce 2009 services, marketing and promotion services, customer relationship management services, and mobile commerce for physical products are not included in the definition of mobile content adopted in this research. It is also important to highlight how specific attributes of mobility relate to the concept of mobile digital content. Clarke (2001) mentions four mobile value proposition attributes: ubiquity, convenience, localization and personalization. By ubiquity, the author indicates the fact that most mobile devices are constantly connected to the network and, as such, are constantly capable of sending and receiving data, virtually meaning availability at any time and everywhere. Similarly, convenience means that mobile users are not restricted by usual time and place constraints, while the localization attribute indicates the ability to easily locate and identify the mobile user. The combined effect of these attributes is that of enabling both proactive pushing and enrichment of mobile content by adding real-time and/or location-enhanced characteristics. Finally, by personalization Clarke (2001) means the fact that mobile phones are extremely personal devices, usually directly linked to only one user, reflecting his own preferences and desires for selfexpression. A possible effect of the personalization attribute is an increased importance of mobile content that can be used to turn the device into a tool for expressing the user's personality and lifestyle. At the same time, it may open up opportunities for enriching informational content by tailoring it to each user own personal characteristics and interests. Considering the particularities of mobile content, it can be concluded that mobile content types are extremely varied and diverse. In fact, many mobile content taxonomies were already proposed (Steinbock, 2005; Netsize, 2007; Netsize, 2008; Holden, 2008a, 2008b, 2008c). Most can be narrowed down to the following types of content, based on the creative or informative value of the content itself (Feijóo et al., 2009): Mobile Television (referring to broadcasting delivery mode) and Video (on-demand delivery); Mobile Music (ringtones and fulltracks); Mobile Gaming; Mobile Adult; Mobile Personalization (wallpapers and images); Mobile User-Generated Content; Mobile Publishing; Mobile Advertising; and Mobile Gambling. It is clear that there is significant overlapping among categories in this taxonomy. For instance, an image may be classified both as Personalization and Adult, while a ringtone may be classified as both Music and Personalization, and User-Generated Content can be classified as almost any other content type. In order to avoid this kind of pitfall, another possibility would be to classify mobile content according to the perceived value for the users (or the value proposed by the offerer) instead of the intrinsic content or informational value itself. This is the approach adopted in this paper, where the following content types are considered: • Infotainment: includes both information-based and entertainment-based content. Examples of information-based content are news accessible through WAP sites or delivered through text alerts, while examples of entertainment-based content include music and videos, both downloaded and streamed, and television broadcasted to mobile phones. • Communication & Community (C&C): content focusing on or enabling interaction, interactivity and/or collaborative content generation by users, such as chat, social networks and forums. • Customization: content designed to customize the device in order to make it reflect the user's own personality and lifestyle. Examples of customization are ringtones, wallpapers and screensavers. • Betting: content related to gambling, such as lotteries and reverse auctions. The content component of these services is mainly the feedback the user gets when placing the bet or bid, or updates during the course of the game. • Gaming: content based on or enabling interaction with a gaming application, either embedded or accessed through the wireless connection. 2.3 Mobile Telecommunications in Brazil Brazil is the largest and most populous country in South America, with circa 192 million population (International Monetary Fund, 2008a). The country has a strong industrial base and well developed commerce and service sectors, being the largest economy in Latin America. Estimations for 2008 indicate a GDP of U$ 1.66 trillion at current prices (a 5.2% growth) and 5.7% annual inflation (International Monetary Fund, 2008a, 2008b). ICT are relatively pervasive: 98% of the households have access to TV, 89% to radio, 74% to mobile phone, 45% to fixed phone and 24% to desktop computer (Balboni, 2008). Brazil has a high Internet diffusion rate when compared to other developing countries: 17% of the households have Internet access (circa half of those broadband), with Internet users reaching 45% of the population (Balboni, 2008). 133 ISBN: 978-972-8924-89-8 © 2009 IADIS At the end of 2008, mobile telephony coverage reached circa 93% of the population, mobile subscribers amounted to 150.6 million, a 24.5% year-over-year increase, and mobile penetration was 78.11% (Teleco, 2009a). These numbers make the country the 5th largest global mobile market by number of subscribers. Moreover, Balboni (2008) found that in 2007 the percentage of the population that owned a mobile phone was 51% (a 9 p.p. increase over 2006), while 66% of the population had used a mobile phone (up from 61% in 2006). National average ARPU was U$ 16.4 at the end of third quarter 2008 (Teleco, 2009c), while the combined gross revenues for the top four MNOs for 2007 was estimated at U$ 31.1 billion (Teleco, 2009b). Subscriber base at the end of 2008 was 80.6% pre-paid, network technology mainly 2G and 2,5G (88.9% GSM; 8.45% CDMA; 0.77% TDMA; 0.01% AMPS), while only 1.87% of the subscribers used 3G technologies (Teleco, 2009a), as it has been commercially introduced only in the course of 2008. Brazil followed the consolidation trend that characterized most mobile markets in the late 90s and early 2000s. As a comparison, while in 2001 circa 40 MNOs operated in Brazil, presently there are only seven, four of them having national coverage (Vivo, Claro, TIM and Oi), while the remaining three (Sercomtel, CTBC and Unicel) having only local coverage. There are no MVNO, as regulation regarding the issue has not yet been defined. It may be argued that MNO are still focused on improving market share, concentrating on new customer acquisition and price competition for voice services. In December 2008, Vivo was the market leader (29.84% market share) followed by Claro and Tim (25.71% and 24.17%, respectively) and Oi (19.91%). Other MNO's market shares are marginal (less than 0.5% combined) (Teleco, 2009c). 3. METHODOLOGY The research is of an exploratory and descriptive nature, as it aims to present a first investigative approach to a complex and contemporary subject. The empirical study had two sequential phases. The first one was quantitative, having as main data collection procedure a Web and WAP-based survey of the mobile content offering in Brazil. A qualitative phase followed in the form of a deep case study (Yin, 2003) on the Brazilian Mobile Content value network involving the interview of 39 market experts, including executives from major mobile content producers, aggregators and retailers. Secondary qualitative data sources, such as journal articles, conference papers, industry-related news, reports and books complemented both phases. The research was conducted in collaboration with the Politecnico di Milano (Italy). Three researchers were involved in the first phase and two in the second. Research design contemplated two distinct stages for the quantitative phase. The first was conducted at distance, using web search engines and web-based WAP emulators, and took place from August to early September 2008. The second stage, during late September 2008, was conducted in loco with mobile phones. During this second stage, researchers also surveyed mobile content advertisement in traditional media. Even if the survey cannot be said to be throughly complete, it can be assumed that it is representative enough for a valid analysis. The fact that survey results where crosschecked by interviewees renders the assumption even more plausible. For the survey, services were categorized according to type of content, technology platform for content delivery or use, and revenue model. The content types were already mentioned in section 2.2: Infotainment, C&C, Customization, Betting, and Gaming. Technology platforms include SMS, MMS, Download, Streaming, Client-Server Application, Browsing, and Digital TV Broadcast. Finally, revenue model include single-buy, subscription, both single-buy and subscription, and free. When possible, information about the content creator, aggregator and retailer, pricing and number of single items offered were collected. A first sample of 56 potential subjects for the interviews was identified during the survey, consisting of top managers for companies considered medially or highly relevant for the market. Criteria for inclusion in this sample were mainly subjective, involving the number of services offered or supplied by the company, media presence and perceived brand strength (both nationally and internationally), and convenience for procuring the interview. Other than managers, this potential sample included a number of market analysts, academia and media experts on the topic in hand. Again, the selection criteria were subjective, a combination of self-declared experience dealing with the market (for media experts and market analysts) and academic relevance measured by number of published papers (for academics). Of these, 25 semi-structured interviews were conducted between October 2008 and January 2009. Additionally, 23 new subjects were considered following a snowball sampling (Coleman, 1958) after the first round of interviews was finalized. Of these, 14 new interviews were conducted during the months of January and February 2009. 134 IADIS International Conference e-Commerce 2009 4. RESULTS A total of 3,678 services from 291 retailers were surveyed, and 39 semi-structured interviews with executives and market experts were conducted. 4.1 Type of Content The most common mobile content type in Brazil is Infotainment (79%), but Customization is also relevant (11%). Interviewees hinted that the high proportion of Infotainment is related to a strong presence of media and web companies playing the role of content retailers and providers. On the other hand, the relatively small presence of companies focusing purely on mobile content retailing may contribute to the lesser role of Customization, which is traditionally the mainstay of this type of player. Moreover, experts mentioned that Customization content suffers from competition with free content available in the Internet. Gaming and C&C represent lesser segments (5% and 4%, respectively). However, they deserve praise for content quality and variety. Diverse propositions of gaming content were found, from prize quizzes promoted by media companies available through Browsing and SMS platforms to downloadable games offered mainly by pure content retailers, even if faced by contrary market structural issues such as high prices for data transmission and low diffusion of sophisticated mobile devices. Interviewees mentioned that games are considerably profitable, being fairly resistant to competition with illegal free content. Additionally, a common perception among experts is that C&C is still underdeveloped, but is expected to see a substantial increase in both offer and demand in the near future, as 3G networks are rolled out and more subscribers adopt the new technology. Finally, Betting content amounts to only 1% of the offering, but is deemed by market experts to represent a substantial part of the overall revenues. This is so because of the reverse auction format, publicized by large media companies with national coverage and later adopted by MNOs and other players, and considered to be extremely profitable. However, reverse auctions are under government scrutiny, as some argue that it does not abide to the current legislation for gambling. 4.2 Technological Platform The leading technological platform for mobile content in Brazil is Browsing, accounting for 40% of the analyzed offering. SMS and Download can also be considered very relevant, representing 32% and 23% of the offering, respectively. Streamed mobile content is incipient, as only 4% of the offering was available through Streaming. Other technological platforms, such as Client/Server Applications (0,3%), MMS (0,5%) and Digital TV Broadcast (1%) are marginal. The overall picture for technological platforms indicates a strong tendency for content delivery through unsophisticated technologies, mainly text messaging. This can be explained by the wide diffusion of technologically limited devices and users being familiar with the SMS technology because of its ample use as alternative to voice calls for personal communication. Interviewees confirmed that although it is not the most prevalent in terms of offering quantity, SMS is the platform that generates most revenues. The limited role of Download may be related to confusing pricing policies and high prices charged by MNOs for data transmission. These barriers, according to the interviewees, are also holding back the expansion of more sophisticated technological platforms, such as MMS and Digital TV Broadcast. The later, besides, mirrors the situation of the general Digital TV technology in Brazil, introduced recently and lacking overall diffusion even in the pure TV format. At the same time, however, the high prices and confusing pricing policies for data transmission may seem to contrast with the significant Browsing offer. Interviews with market experts helped clarify this apparent paradox. According to their perception, mobile users in Brazil either do not access WAP at all or are moderately heavy users. Although a decisive minority (estimates vary from 3 to 5 million), these heavy users are highly profitable, as they usually own high-end mobile devices, know beforehand what kind of content they want and how to obtain it, and make use of discounted data plans. Moreover, experts believe that heavy WAP users tend to be experienced Internet users, and expect a comparable level of content quality in the WAP platform. Interviewees also indicated that some content offerers prefer the WAP instead of the Web as retail channel in order to avoid direct competition with illegal content sources, which in their perception tend 135 ISBN: 978-972-8924-89-8 © 2009 IADIS to be sought after by Brazilian web users. Finally, some market experts pointed out the fact that MNO invested heavily in the WAP platform in the early 2000s, expecting substantial returns in the form of data revenues and favoring Infotainment WAP sites open to consultation without premium charges. Indeed, 90% of the Browsing offer is composed by Infotainment. 4.3 Revenue Model Premium mobile content in Brazil comprises 58% of the surveyed offering. It is important to notice the relevance of the subscription revenue model, as more than half of the premium mobile content can be subscribed (47% requires subscription, while 11% can be acquired either by subscription or single buy). Interviews indicated that content retailers are presently changing their single-buy offer to subscription-based revenue models, in order to stabilize a demand that otherwise can be very unpredictable. Market experts interviewed were convinced that single buy revenue model will markedly decrease in the near future. The complementary free content (42% of the overall offering) mostly represents Infotainment available through Browsing. Indeed, 89% of this free content was classified as Infotainment, and included WAP sites with rich and relevant content offered by renowned players from traditional media and web sectors. Experts pointed out a revenue model where MNOs and content provider share revenues from data traffic in content provider's WAP site as an important driver for this type of offering. 4.4 Transversal Issues Besides providing clarification about the offering, interviews highlighted transversal issues related to structural aspects of the market. A first issue is that of the high prices for data transmission, already mentioned before. Coupled with the usual complex and confusing pricing policies for data transmission, this is deemed by market experts as an important barrier to mobile content diffusion in Brazil. It is believed that MNOs should address this barrier in the near future in order to stimulate market development. A second issue unanimously cited as one of the main barriers for market growth is that of the high taxes. Interviewees complain that effective taxes can reach as much as 35% of the gross revenues. Moreover, they pointed out the fact that some taxes are collected at each level of the value network, resulting in multiple taxation for the same transaction. A third issue mentioned was the lack of market regulation by the government, and players themselves have not reached a consensus for auto-regulation. In the meanwhile, a growing number of consumer post complains regarding bad market practices, such as confusing and misleading subscription models. Finally, a fourth transversal issue relates to the relationship between MNOs and the rest of the value network. Having ownership over the end customer, MNOs have the strongest bargaining power in the whole value network. As a result, revenue share agreements may not generate enough remuneration to the middle and upper tiers. In fact, some market experts suggested that the current share of revenue retained by most MNOs, circa 50% of the gross revenue, can be a significant barrier to further development of the mobile content market, as it obviates experimental retail models or value propositions that do not rely on large volumes of sales. On the other hand, MNOs' stronger bargaining power also means that they may impose harsh conditions for new entrants, or even for incumbents that wish to launch new services or experiment with new revenue and value proposition formats. Finally, some experts pointed out that MNOs do not see mobile data as strategically important, as there is still much competition going on for market share in the voice market. As a result, experts mentioned that investments in data infrastructure by the MNOs lagged in the past. Although this is rapidly changing with the advent of the 3G networks, some experts still believe MNOs' data resources to be inefficient, particularly the billing systems. This, combined with low ARPU, results in extremely high billing failure levels, which further discourages investments in mobile content. 5. CONCLUSIONS In the light of the growing importance of data services in mobile telecommunications, this study presented an overview of the Brazilian mobile content market. A two-phase mixed methods exploratory research was conducted. The first phase consisted of a web and WAP-based census of 3678 mobile content services. It was 136 IADIS International Conference e-Commerce 2009 complemented by a second, qualitative phase made up by a deep case study about the Brazilian Mobile Content value network consisting of 39 semi-structured interviews with key executives and experts. Results indicate that mobile content in Brazil is extremely rich and diverse, although technologically restricted by the fact that most mobile devices in the market have limited technological capabilities and 3G networks are recent. Infotainment is the predominant offering, while Customization lags behind. Gaming and C&C content, although not much representative, are considered to have remarkable quality. In particular, C&C is believed to have high growth potential in the near future. Similarly, SMS and Browsing platforms were found to be especially relevant. While SMS is popular because of its simplicity and low cost, quality WAP sites are a direct result of investments by the MNOs in the early days of WAP combined with a perceived profitability related to WAP users and a remuneration model that shares data transmission revenues between MNO and content provider. Low average ARPU, high taxes, high prices charged by MNO for data traffic and confusing pricing policies were identified as significant barriers for the further development of the mobile content market in Brazil. A number of limitations to the study must be highlighted. First of all, the mobile content market is a new and dynamic business environment, subject to constant and rapid change. This, coupled with limitations derived from a few methodological choices, make it so that the offering mapped cannot be said to fully represent the complete mobile content offer in Brazil. Even so, it can be argued to be sufficiently representative of that particular instant, and sufficient for the analysis desired. Furthermore, the lack of comparative data renders impossible to establish an evolutionary profile for the market. Finally, a third limitation may be established by the nature of the research itself. As an exploratory and mainly descriptive study, the present research is somewhat limited in its power to point out eventual correlation effects and draw explanatory conclusions. These should be the object of further investigations, which may build up on the initial overview provided by this paper. REFERENCES Balboni, M., 2008. Survey on the use of Information and Communication Technologies in Brazil: ICT Households and ICT Enterprises 2007. Comitê Gestor da Internet no Brasil, São Paulo, Brasil. Barnes, S.J., 2002. The mobile commerce value chain: analysis and future developments. International Journal of Information Management, Vol. 22, No. 2, pp. 91-108. Barnes, S.J. & Scornavacca, E., 2007. The emergence of mobile commerce. In S.J. Barnes (editor), E-Commerce and VBusiness: Digital Enterprise in the Twenty-First Century, 2nd edition. Butterworth-Heinemann, Oxford, UK. Bouwman, H. et al., 2008. Trends in mobile services in Finland 2004-2006: from ringtones to mobile Internet. Info, Vol. 10, No. 2, pp. 75-93. Clarke, I., 2001. Emerging value propositions for m-commerce. Journal of Business Strategies, Vol. 18, No. 2, pp. 133148. Coleman, J.S., 1958. Relational Analysis: The Study of Social Organization with Survey Methods. Human Organization, Vol. 17, No. 4, pp. 28-36. Dunnewijk, T. & Hultén, S., 2007. A brief history of mobile communication in Europe. Telematics and Informatics, Vol. 24, No. 3, pp. 164-179. Feijóo, C. et al., 2009. Exploring a heterogeneous and fragmented digital ecosystem: Mobile content. Telematics and Informatics, Vol. 26, No. 3, pp. 282;292. Frost & Sullivan, 2008. Latin American Mobile Content Markets, retrieved February 2, 2009 from http://www.researchandmarkets.com/reports/612484. Funk, J.L., 2006. The future of mobile phone-based Internet applications: A view from Japan. Technovation, Vol. 26, No. 12, pp. 1337-1346. Funk, J.L., 2007. Solving the startup problem in Western mobile Internet markets. Telecommunications Policy, Vol. 31, No. 1, pp. 14-30. Funk, J.L., 2009. The emerging value network in the mobile phone industry: The case of Japan and its implications for the rest of the world. Telecommunications Policy, Vol. 33, No. 1-2, pp. 4-18. Holden, W., 2008a. Mobile Entertainment Markets – Opportunities & Forecasts 2007-2012. Juniper Research Limited, Hampshire, England, 2008a. Holden, W., 2008b. Mobile Content Strategies & Business Models – Scenarios & Forecasts 2008-2013. Juniper Research Limited, Hampshire, England. 137 ISBN: 978-972-8924-89-8 © 2009 IADIS Holden, W., 2008c. Mobile Gambling: Casinos, Lotteries & Betting 2008-2013, 5th Edition. Juniper Research Limited, Hampshire, England. International Monetary Fund, 2008a. World Economic Outlook Database, October 2008, retrieved February 1, 2009 from http://www.imf.org/external/pubs/ft/weo/2008/02/weodata/index.aspx. International Monetary Fund, 2008b. World Economic Outlook. International Monetary Fund, Washington, DC, USA. Kallio, J. et al., 2006. An international comparison of operator-driven business models. Business Process Management Journal, Vol. 12, No. 3, pp. 281-298. Kitson, A., 2008. BRIC Mobile Opportunities – Markets & Forecasts 2008-2013, Juniper Research Limited, Hampshire, England. Loebbecke, C. & Huyskens, C., 2007. Online delivered content: concept and business potential. In S.J. Barnes (editor), ECommerce and V-Business: Digital Enterprise in the Twenty-First Century, 2nd edition. Butterworth-Heinemann, Oxford, UK. Lu, Y. et al., 2007. The mobile business value chain in China: a case study. International Journal of Electronic Business, Vol. 5, No. 5, pp. 460-477. Maitland, C.F. et al., 2002. The European market for mobile data: evolving value chains and industry structures. Telecommunications Policy, Vol. 26, No. 9, pp. 485-504. Michael, A. & Salter, B., 2006. Mobile Marketing: Achieving Competitive Advantage Through Wireless Technology. Butterworth-Heinemann, Oxford, UK, 2006. Mika, N., 2009. Mobile connections surpass 4 billion mark worldwide. Reuters Technology, retrieved February 9, 2009 from http://www.reuters.com/article/technologyNews/idUSTRE51A2I820090211. Netsize, 2007. The Netsize Guide – Convergence: Everything is going mobile. Netsize, Paris, France. Netsize, 2008. The Netsize Guide – Mobile 2.0, you are in control. Netsize, Paris, France. O'Donnel, J. et al., 2007. Australian Case Studies in Mobile Commerce. Electronic Commerce Research, Vol. 2, No. 2, pp. 1-18. Peppard, J. & Rylander, A., 2006. From Value Chain to Value Networks: Insights for Mobile Operators. European Management Journal, Vol. 24, No. 2-3, pp. 128-141. Richtel, M., 2009. Can the Cellphone Industry Keep Growing? The New York Times, retrieved 5 February, 2009 from http://www.nytimes.com/2009/02/04/technology/companies/04cell.html. Scornavacca, E. et al., 2006. Mobile business research published in 2000-2004: emergence, current status, and future opportunities, Communications of the Association for Information Systems, Vol. 17, pp. 635-646. Steinbock, D., 2005. The Mobile Revolution – The Making of Mobile Services Worldwide. Kogan-Page, London. Teleco, 2009a. Estatísticas de Celular no Brasil, retrieved February 5 2009 from http://www.teleco.com.br/ncel.asp. Teleco, 2009b. Operadoras de Celular no Brasil: Econômico Financeiro, retrieved February 5 2009 from http://www.teleco.com.br/opcelular.asp. Teleco, 2009c. Operadoras de Celular no Brasil: Operacional, retrieved February 5, 2009 from http://www.teleco.com.br/opcelular.asp. Teleco, 2009d. Operadoras de Celular no Brasil: Receita de Dados, retrieved February 5 2009 from http://www.teleco.com.br/opcelular_dados.asp. The Insight Research Corporation, 2007. The 2008 Telecommunications Industry Review – An Anthology of Market Facts and Forecasts. The Insight Research Company, Boonton, New Jersey. Yin, R., 2003. Case study research: Design and methods. Sage Publishing: Thousand Oaks, CA, USA. 138 IADIS International Conference e-Commerce 2009 BUYER’S MINDSET ABOUT ONLINE PURCHASE AND AUCTION, AND ITS EFFECT ON PAYMENT CHOICE Wee Kheng Tan Assistant Professor, Department of Information & Electronic Commerce, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County, Taiwan Yung Lun Chung Department of Information & Electronic Commerce, Kainan University, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County, Taiwan ABSTRACT Payment is an inborn part in any e-transaction. Accompanied with any e-transaction is the idea of risk. This risk is more prevalent in e-auction than e-purchasing because for the former, the transaction parties are individuals and e-auction fraud is often being heard of. Different payment methods are in used or developed to address such fear. Cash, credit card and ATM transfer are some of the payment method used by online buyers. Another major trend is the intertwining of cash and material flow, usually with payment being made at the same place and time the buyer gets the product. This research uses two types of online shopping, e-purchase (B2C) and e-auction (C2C) as the subjects of investigation. A set of criteria is also derived to account for the fact that buyers will consider many criteria to reach a decision on using which payment method. Since no single criteria can adequately explain the choice of buyers, Analytical Hierarchy Process (AHP) method is used for analysis. This research shows that buyers’ decision on the method of payment is a multicriteria consideration and the relative importance of the criteria reflects the nature of the transaction. From the buyer’s perspective, making payment always involve the risk of not receiving the product or the product is not as expected. Furthermore, the risk associated with online auction is more than online purchasing. As a result, buyers hope to lower the risk of the transaction by choosing payment method that safeguards their interest. Hence, risk-related criteria such as secure payment method and product’s price level are top on the list of importance and such consideration is regardless of the nature of transaction. However, such concern is moderated if the payment method is already widely used by buyers. KEYWORDS e-Purchase, e-Auction, Payment Method, AHP Method. 1. INTRODUCTION As more and more Taiwanese are attracted by the lower prices and abundant product choices in online shopping sites, online shopping is getting increasingly popular in Taiwan. According to Market Intelligence and Consulting Institute, online sales in Taiwan was estimated to be NT$243 (US$7.32) billion in 2008 with B2C (online auction) transactions and C2C transactions accounting for NT$136 (US$4.10) billion and NT$107 (US$3.22) billion respectively (Taipei Times, 2009). Associated with any e-transaction are material flow, cash flow and information flow with payment being an inborn part in any e-transaction (Tsiakis and Sthephanides, 2005). Accompanied with e-transaction is the idea of risk. The buyers are concerned that after payment has been made, the product ordered never arrived or is not according to specification. For the sellers, they are worried that payment never arrived even though the product has been delivered. The fear is even more prevalent for e-auction than for e-purchasing because for the former, the transaction parties are individuals that can be hidden being the anonymity of Internet and eauction fraud is often heard of. Different payment methods are being used or developed, though with various extent, to address such risk and fear. Cash payment, credit card payment and ATM transfer are some of the traditional payment methods used by buyers of online purchase and auction website. PayPal, a form of C2C intermediary payment service, provides a secure payment environment for payment. Another major trend is the intertwining of cash and 139 ISBN: 978-972-8924-89-8 © 2009 IADIS material flow, usually with payment being made at the same time the buyers get their products. In Taiwan, emerchants co-operate with convenience store operators to allow their buyers to collect the merchandises at the convenience stores specified by the buyers and to collect payment from the buyers. In the past few years, Taiwan has also seen rapid growth in courier companies offering delivery service for B2C and C2C sellers. They may also collect payment on behalf of the sellers. In view of the above, this research used two types of online shopping, e-purchase (B2C) and e-auction (C2C) as the subjects of investigation and for comparison purpose. These two types of e-shopping have different perceived risk level. Using the payment methods chosen through literature review and advices from experts, buyers and sellers, we investigated how the concern for risk would affect the choice of payment method. We also devised a set of criteria to account for the fact that buyers would consider many criteria before reaching a decision on using which payment method. Since these criteria were not necessarily of equal importance and no single criteria could adequately explain the choice of buyers, Analytical Hierarchy Process (AHP) method was used for analysis. This research attempted to answer the following questions: were there significant differences in consideration when buyers of e-purchase and e-auction considered which payment method to use and how these differences ultimately affected their choice of payment method. 2. RESEARCH METHOD Analytic Hierarchy Process (AHP) was first proposed by Thomas L. Saaty (Saaty, 1980) and its main characteristic is based on pair-wise comparison judgments. It is often used to analyze complex multi-criteria decision-making problem and applied to a wide range of problems and applications. The problem investigated is set up as a sequential step-up hierarchy structure with different measures being integrated into a single overall score for ranking decision alternatives. When using AHP, the problem’s step-up hierarchy structure need to be developed. A hierarchy is a specific type of system, based on the assumption that the entities can be grouped into disjoint sets, with the entities of one group influencing those of other groups (Saaty, 1980). Pair-wise comparison matrix forms an important component of AHP. Two criteria are compared using a nine-point scale, where ‘1’ denotes ‘equal’ importance, ‘3’ represents ‘slightly’ importance, ‘5’ indicates ‘clearly’ superiority, ‘7’ is ‘very’ important and ‘9’ denotes ‘extremely’ important, with the even numbers being used to indicate intermediate values if necessary. If there are n criteria to consider, n(n-1)/2 pair-wise comparisons would need to be made. Afterwhich, reciprocal n x n matrix are constructed and weightings are then obtained. The consistency of the pair-wise comparison matrix needs to be verified and the following two indexes can be used: Consistency Index (C.I.) and Consistency Ratio (C.R.). They are defined as in equation (1) and (2) with λmax being the largest eigenvalue and R.I. (Random Index) is as shown in Table 1. For consistency, C.I and C.R. should be less than 0.1 for the AHP analysis to be acceptable. C.I. = ( λmax – n ) / ( n – 1) (1) C.R. = C.I. / R.I. (2) Table 1. Random Index (R.I.) n R.I. 1 0 2 0 3 0.58 4 0.90 5 1.12 6 1.24 7 1.32 8 1.41 9 1.45 10 1.49 3. RESEARCH FRAMEWORK The relevant literature was used as a starting point to identify the key criteria suitable for this research. Instead of just focusing on issues relevant to the payment method itself, we believe that selection of payment method will also depend on the characteristics of the buyers themselves as well as the nature of the products they are buying. Discussions with researchers, buyers at purchase website and auction website were then made to refine the criteria. Two four-tiered hierarchical structures, one each for purchase website and auction website were obtained and as shown in Figure 1 and 2. 140 IADIS International Conference e-Commerce 2009 Figure 1. The Hierarchical Structure for Purchase Website Figure 2. The Hierarchical Structure for Auction Website The second tier for both purchase and auction website comprises three aspects, which are firstly, buyer’s characteristics, secondly, characteristics of payment methods and thirdly, product characteristics. For the structure that depicts e-purchase, the third tier comprises 9 criteria and for the structure that depicts e-auction, there are 8 criteria. The payment methods investigated for purchase website are ATM transfer, online credit card payment, payment through courier service and payment at the convenience store. As for the auction website, the payment methods investigated are ATM transfer, cash payment when meeting face-to-face, payment through courier service and payment through PayPal-type of service. There are two criteria under the aspect of buyer’s characteristics: past experience of e-purchase/e-auction and payment method which buyer is already used to. According to Schmid (1994), buyers could be classified based on the number of time they patronize i.e. first-time buyers, second-time buyers and multi-buyers or advocates. As first-time buyers, they are at the exploratory stage and confirming or de-confirming their expectation so that they could decide whether to re-patronize i.e. became second time buyers. Furthermore, Dawson, Bloch and Ridgway (1990) showed that transient emotions influenced shoppers’ satisfaction and future shopping intentions. Habit can also affect the preference of buyers over the payment method they are using. Habit is the “learned sequences of acts that become automatic responses to specific situations which may be functional in obtaining certain goals or end states” (Verplanken et al., 1998) and the action requires little conscious attention and only minimal mental effort (Ouelette and Wood, 1998). As for e-purchase, the four criteria under the aspect of characteristics of payment method are payment method which is acceptable to e-merchant, easy and simple to use procedure and user interface, secure 141 ISBN: 978-972-8924-89-8 © 2009 IADIS payment procedure, and lastly bonus, discount and other financial benefits. As for e-auction, the e-merchant of the first criteria is replaced by seller and the last criterion for e-purchase is not applicable. The popularity of a product or service is dependent on whether it is easy to use i.e. ease of use. An important measure of the Technology Acceptance Model (TAM), ease of use is defined as ‘the degree to which a person believes that using a particular system would be free from effort’ (Davis, 1989). This research applied ease of use to the payment mechanism. Risk may be defined as uncertainty concerning the occurrence of a loss (Rejda, 2008). It is closely related to the degree a particular product or service is being adopted; the higher the risk, the greater the reluctance for one to adopt the product (Pamela and John, 1998). Transaction via Internet is accompanied by its own set of risks, some of them are applicable to all types of transactions, whether online or offline, while some are specific to the online environment. However, if the buyers want to purchase an item online, they must accept the presence of such risk and if wanted to, use other mechanisms to moderate it. Careful selection of payment method is one of the methods. On the other hand, payment methods themselves also introduce their own set of risk e.g. credit card fraud and malicious tapping of personal information. Hence security of the payment mechanism is a major aspect that buyers need to consider. The last three criteria are related to product’s characteristics: price level, whether attributes of products are easy to ascertain and whether the product is in digital form. In the computer-mediated environment, it is difficult for buyers to examine the product before purchasing. Hence the risk of buying misrepresented products is increased. As a result, buyers often have to factor in the attributes of the products prior to purchasing and it may have an impact on the payment method. Products may be classified in different ways. Degeratu et al. (2000) proposed two categories: sensory products and non-sensory products. Lal and Sarvary (1999) defined two types of product attributes: digital attributes, which can be easily communicated on the Internet, and non-digital attributes. Figueiredo (2000) suggested an e-commerce product continuum in which products are characterized as commodity products (e.g., oil, shares), quasi-commodity products (e.g., books, new cars), look and feel products (e.g., cosmetics, homes), and look and feel products with variable quality (e.g. works of art, used cars). The quality of commodity products is the easiest to ascertain, hence buyers are most confidence of their quality. On the other hand, the quality of the look and feel products with variable quality is the most difficult to ascertain and buyers are least confidence of their quality prior to consuming them. For some of the buyers, a higher price may indicate high quality. However, from the perspective of cash flow, a higher price will also imply a higher risk in the transaction, resulting in more careful consideration of the payment method to lower the risk of the transaction. 4. RESULT AND DISCUSSION As it was not an easy task to fill up the AHP questionnaire, face-to-face survey was arranged to increase the accuracy of the returns. The survey lasted for 2.5 months starting from December 1, 2008. A total of 57 valid returns were received from buyers of purchase websites and 63 valid returns were obtained from buyers of auction websites. Their profile is as shown in Table 2. Table 2. Profile of Survey Respondents Gender Age Monthly Income Engage in e-Purchase/e-Auction Activity Male Female < 35 year old ≥ 35 year old < NT$30,000 ≥ NT$30,000 < 18 months ≥ 18 months e-Purchase No Percent 33 57.9 24 42.1 46 80.7 11 19.3 44 77.2 13 22.8 22 38.6 35 61.4 e-Auction No Percent 34 54.0 29 46.0 50 79.4 13 20.6 43 68.3 20 31.7 25 39.7 38 60.3 4.1 Purchase Website As observed from Table 3, buyers considered characteristics of payment method (0.3890) to be the most important, then product’s characteristics (0.3387). The least considered is buyer’s characteristics (0.2723). 142 IADIS International Conference e-Commerce 2009 For the criteria under buyer’s characteristics, past experience of e-purchase (0.5582) was ranked first, followed by payment method which buyer is already used to (0.4418). For characteristics of payment method, secure payment procedure (0.4445) was top on the list of importance and its weighting was way ahead of the other criteria under the same aspect. Security being top on the list was not surprising. For product’s characteristics, price level (0.4585) came in first, whether attributes of products are easy to ascertain (0.3162) second. These two criteria indicated the degree of uncertainty and risk of the purchase which would ultimately influenced the chosen payment method. Through the overall ranking analysis, it was observed that the top 5 important criteria were all related to buyer’s perception of risk and response so that buyer’s interest would be most safeguarded. These 5 risk-related criteria were secure payment procedure, price level, past experience of e-purchase, payment method which buyer is already used to, and lastly whether attributes of the product are easy to ascertain. Secure payment procedure was top in the mind of buyers. This result confirmed past research which found security to be very important to the success of payment method. Price level of the product was the second most important. Past experience of e-purchase was more important than payment method which buyer is already used to. A major reason was the seller was not individual but merchant and usually they would offer a range of payment options, hence the former criterion was more important. After the risk and associated responses were taken into account, bonus, discount and other financial benefits (ranked 6th), easy and simple to use procedure and user interface (ranked 8th) began to appear. Table 3. Weightings of e-Purchase Aspects and Criteria Aspect 0.2723 (3) Weighting (Ranking) Criteria A. Buyer’s Characteristics A1. Past Experience of e-Purchase A2. Payment Method Which Buyer is Already Used To For Criteria related to Buyer’s Characteristics : C.I. = 0.000；C.R. = 0.000 B. Characteristics of Payment Method 0.3890 (1) B1. Payment Method Which is Acceptable to e-Merchant B2. Easy and Simple to Use Procedure and User Interface B3. Secure Payment Procedure B4. Bonus, Discount and Other Financial Benefits For Criteria related to Characteristics of Payment Method: C.I. = 0.022；C.R. = 0.024 C. Product’s Characteristics 0.3387 (2) C1.Price Level C2.Whether Attributes of Products are Easy to Ascertain C3.Whether the Product is in Digital Form For Criteria related to Product’s Characteristics: C.I. = 0.001；C.R. = 0.001 For Aspect: C.I. = 0.003；C.R. = 0.005 Overall 0.5582 (1) 0.4418 (2) 0.1520 (3) 0.1203 (4) 0.1004 (4) 0.1896 (3) 0.4445 (1) 0.2655 (2) 0.0391 (9) 0.0738 (8) 0.1729 (1) 0.1033 (6) 0.4585 (1) 0.3162 (2) 0.2253 (3) 0.1553 (2) 0.1071 (5) 0.0762 (7) As shown in Table 4, payment at convenience store (0.2687) was the most preferred method of payment for the e-purchase buyers. This outcome was a combination of various reasons. Firstly, the buyer could gain immediate possession of the product when payment was made, hence removing the concern about making payment but not receiving the product. The analysis revealed that this method had very high weighting for risk-related criteria, especially secure payment procedure and price level. Secondly, this payment method was already a familiar and popular method among existing buyers, as evidenced by the high weighting of payment method which buyer was already used. Its familiarity was only lower than online credit card payment. This reason was a result of the shopping habit and retail structure of Taiwan. Many Taiwanese were very used to making purchase at the convenience stores and patronizing these outlets was almost a daily affair event for them. Taiwan had the highest density of convenience stores in the world with 9,100 convenience stores in 2009 for an island, with an area of 36,000 sq km and a population of about 23 million (CENS.com, 2009). Hence, payment at convenience store offered the most ‘peace of mind’ option. Online credit card payment (0.2662) was the next favored payment method. Online credit card payment was ranked as the second payment method which buyer was already most used to and first in past experience of epurchase. Taiwan had a high credit card penetration rate with 19.89 million effective credit cards in use in September, 2008 (CENS.com, 2008). Many Taiwanese were used to paying through credit card though there were on-and-off complaints about credit card fraud. Credit card fared well in security. Familiarity with credit 143 ISBN: 978-972-8924-89-8 © 2009 IADIS card did help and it was also an endorsement by buyers of the efforts made by merchants and credit card issuers to provide a secure environment. The post-payment method also contributed to its popularity and the discount and rebates further sweetened the deal. Making payment when courier delivered the product (0.2831) offered almost the same ‘peace of mind’ to the buyers as making payment at convenience store, however, it performed poorer in most other criteria when compared to payment at convenience store and online credit card payment. ATM transfer (0.2270) came in last because it offered the least protection to the buyers. Table 4. Weightings of Methods of Payment for e-Purchasing Payment Method Payment at Convenience Store Online Credit Card Payment Payment through Courier Service ATM Transfer Weighting (Ranking) 0.2687 (1) 0.2662 (2) 0.2381 (3) 0.2270 (4) 4.2 Auction Website As observed from Table 5, buyers considered product’s characteristics (0.3974) to be the most important aspect, followed by characteristics of payment method (0.3243). The least considered was buyer’s characteristics (0.2783). For the criteria under buyer’s characteristics, payment method which buyer is already used to (0.5757) was ranked first, followed by past experience of e-purchase (0.4243). For characteristics of payment method, secure payment procedure (0.5723) was top on the list of importance and its weighting was way ahead of the rest of the criteria under the same aspect. For the criteria under product’s characteristics, price level (0.4672) came in first, followed by whether attributes of products are easy to ascertain (0.3604). These two criteria indicated the high degree of uncertainty and risk of e-auction which would ultimately affected the choice of payment method. The overall ranking analysis revealed that the top 5 important criteria were all related to buyer’s perception of risk and response so that buyer’s interest would be most safeguarded. Price level (0.1857) was top in the mind of buyers. Secure payment procedure (0.1856) was the second most important. Payment method which buyer is already used to (0.1602) came in third, reflecting the careful and conservative nature of buyers. The other two top 5 important criteria were whether attributes of products are easy to ascertain (ranked 4th) and past experience of e-auction (ranked 5th). After the risk and associated responses were taken into account, easy and simple to use procedure and user interface (ranked 6th), whether the product is in digital form (ranked 7th) and payment method which is acceptable to seller (ranked 8th) began to appear. Table 5. Weightings of e-Auction Aspects and Criteria Aspect 0.2783 (3) Weighting (Ranking) Criteria A. Buyer’s Characteristics A1. Past Experience of e-Auction A2. Payment Method Which Buyer is Already Used To For Criteria related to Buyer’s Characteristics : C.I. = 0.000；C.R. = 0.000 B. Characteristics of Payment Method 0.3243 (2) B1. Payment Method Which is Acceptable to Seller B2. Easy and Simple to Use Procedure and User Interface B3. Secure Payment Procedure For Criteria related to Characteristics of Payment Method: C.I. = 0.007；C.R. = 0.013 C. Product’s Characteristics 0.3974 (1) C1.Price Level C2.Whether Attributes of Products are Easy to Ascertain C3.Whether the Product is in Digital Form For Criteria related to Product’s Characteristics: C.I. = 0.003；C.R .= 0.004 For Aspect: C.I. = 0.013；C.R. = 0.023 Overall 0.4243 (2) 0.5757 (1) 0.1181 (5) 0.1602 (3) 0.1606 (3) 0.2671 (2) 0.5723 (1) 0.0521 (8) 0.0866 (6) 0.1856 (2) 0.4672 (1) 0.3604 (2) 0.1724 (3) 0.1857 (1) 0.1432 (4) 0.0685 (7) As shown in Table 6, payment through courier service (0.2964) presented the best option for buyers because it offered the best combination of lowest risk and convenience when compared to other two top-three 144 IADIS International Conference e-Commerce 2009 payment methods: ATM transfer and cash payment when meeting face-to-face. Payment through courier service allowed buyers to pay only when they received the product at a location specified by them. It was evidenced by the highest weighting for secure payment procedure and price level. It also performed very well for other risk-related criteria. Preference for ATM transfer was a result of its popularity as a payment method. ATM transfer performed very well for past experience of e-auction and especially for payment method which buyer was already used to. It reflected a compromise between the concern of the buyers and sellers. It was hence not surprising for ATM transfer to be top in the weighting for method which was acceptable to seller. Even though ATM transfer might be seen to favor the seller slightly more, however given its popularity, the comfort level and the ‘peace of mind’ level perceived by the buyers might not be as low as one might think. Cash payment when meeting face-to-face guaranteed ‘peace of mind’ for both buyers and sellers, nevertheless both parties needed to arrange a place and time to meet, which might be troublesome for some. As for PayPal-type of service (0.1575), the ranking was the lowest, indicating the conservative nature of the buyers and in addition, once certain payment method was well established and accepted, introducing new payment method and changing the mind of the users would not be that easy. Seller not asking for this kind of payment method was also one of the contributory reasons. Table 6. Weightings of Methods of Payment for e-Auction Payment Method Payment through Courier Service ATM Transfer Cash Payment when Meeting Face-to-face Payment through PayPal-type of Service Weighting (Ranking) 0.2964 (1) 0.2872 (2) 0.2589 (3) 0.1575 (4) 4.3 Overall Discussion Even though it is not possible to compare the weighing of e-purchase and e-auction directly because of the presence of one additional criterion for e-purchase, this research is still able to obtain some insights of the buyer’s mindset and the consequent effect on their choice of payment method. Buyers already have some pre-conceived idea about e-purchase and e-auction. This mindset exists because of the difference in the nature of e-purchase and e-auction. As a form of C2C e-commerce, both the buyer and seller of e-auction website are individuals. With Internet auction fraud being a common feature, coupled with lack of face-to-face interaction and difficulty in ascertaining the identity of both parties, concern about fraud is high. Even though the monetary amount per e-auction transaction may be low, it does not lower such concern since the transaction parties are individuals. For online purchase website, the seller is now a business entity. The risk perceived by e-purchase buyer will be lower but not eliminated as there is still concern of bogus merchant (which can be lowered further by buying from reputable merchants) and products which do not meet expectation. Such kind of mindset is clearly manifested through the weightings obtained. The top 5 criteria in the overall ranking for e-purchase and e-auction are all related to buyers’ perception of the risk and their response. The two criteria from buyer’s characteristics are past experience of epurchase/e-auction and payment method which buyer is already used to. The next two criteria, price level and whether attributes of products are easy to ascertain are from product characteristics. The only criterion from characteristics of payment method is secure payment procedure. It is only after these risk-related criteria are considered, other criteria less related to risk such as ease of use, whether product is in digital form and acceptable to seller surface. Hence, this research concludes that risks associated with the transaction and payment method is paramount to all parties when deciding on the payment method. At the level of aspects, e-auction buyers are concerned firstly about product’s characteristics, followed by characteristics of payment method. However, for e-purchase buyers, it is the other way round. The key criteria under product’s characteristics which are of concern to all parties are price level and whether attributes of products are easy to ascertain. As for characteristics of payment method, it is secure payment procedure. This outcome indicates that e-auction buyers are more concerned than e-purchase buyers about the risk associated with the product itself. The difference in emphasis on product’s characteristics and characteristics of payment method coupled with buyer’s characteristics has a major impact on the option of payment method. Buyers of e-auction website prefer payment through courier service, followed by ATM transfer, cash payment when meeting face-to-face and lastly payment through PayPal-type of service. Buyers of e-purchase website prefer payment 145 ISBN: 978-972-8924-89-8 © 2009 IADIS at convenience store, online credit card payment, payment through courier service and lastly ATM transfer. Given the risk involved, both types of buyers clearly show preference for payment method which offers them the greatest degree of security, i.e. payment through courier service for e-auction buyers and payment at convenience store for e-purchase buyers. These two methods allow buyers to receive their products and make payment simultaneously. However, maximizing their security is not a hard and fast rule. Buyers, when needed to, do take into account payment methods which are already popular among the buyers, such as credit card and ATM transfer. As a proverb says ‘old habit dies hard’, even though ATM transfer offers higher risk for e-auction buyers, it is still the second most popular method. ATM transfer is a common method and old habit and familiarity breed greater degree of comfort among the buyers. 5. CONCLUSION This research shows that buyers’ decision on using which payment method is a multi-criteria consideration and the relative importance of the criteria reflects the nature of the transaction. From the buyer’s perspective, making payment always involve the risk of not receiving the product or the product is not as expected. Furthermore, the risk associated with online auction is more than online purchasing. As a result, buyers hope to lower the risk of the transaction through choosing payment method that best safeguards their interest. Hence, regardless of the nature of transaction, risk-related criteria such as secure payment method and product’s price level are top on the list of importance. However, such concern is moderated if the payment method is already widely used among the buyers. This research also shows that payment at convenience store is a good example of cash flow and material flow being closely intertwined and this payment method also has high popularity in Taiwan. A follow-up would be to conduct a multi-country research to find out whether there is a difference in mindset of the online buyers among different countries. REFERENCES CENS.com, 2008. Taiwan’s Credit Card Market Gaining Strength. Viewed 1 December, 2008, <http://www.cens.com/cens/html/en/news/news_inner_25325.html>. CENS.com, 2009. Growth Slows for Taiwan Convenience Stores. Viewed 6 January, 2009, <http://www.cens.com/cens/html/en/news/news_inner_25813.html>. Davis F. D., 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. In MIS Quarterly, Vol. 13, No. 3, pp. 319-340. Dawson S., Bloch P. H. and Ridgway N. M., 1990. Shopping Motives, Emotional States, and Retail Outcomes. In Journal of Retailing, Vol. 60(Winter), pp. 408–427. Degeratu A., Rangaswamy A. and Wu J., 2000. Consumer Choice Behavior in Online and Traditional Supermarkets: The Effects of Brand Name, Price, and Other Search Attributes. In International Journal of Research in Marketing, Vol. 17, pp. 55-78. Figueiredo J., 2000. Finding Sustainable Profitability in Electronic Commerce. In Sloan Management Review, pp.41-51. Lal R. and Sarvary M., 1999. When and How is the Internet Likely to Decrease Price Competition?, In Marketing Science, Vol. 18, No. 4, pp. 485–503. Ouellette J. A. and Wood W., 1998. Habit and Intention in Everyday Life: The Multiple Processes by which Past Behavior Predicts Future Behavior. In Psychological Bulletin, Vol. 124, No. 1, pp. 54-74. Pamela D. and John H. R., 1998. Matching Electronic Distribution Channels to Product Characteristics: The Role of Congruence in Consideration Set Formation. In Journal of Business Research, Vol. 41, No. 3, pp. 223-229. Rejda G., 2008. Principles of Risk Management and Insurance 10th Edition, Addison Wesley, Boston, USA. Saaty T. L., 1980. The Analytic Hierarchy Process. McGraw-Hill, New York, USA. Schmid S., 1994. Building Loyal Buyers, In Catalog Age, Vol. 11, No. 10, pp. 141-144. Taipei Times, 2009. Brisk Outlook for Online Shopping: Poll. 30 January. Tsiakis T. and Sthephanides G., 2005. The Concept of Security and Trust in Electronic Payments. In Computers & Security, Vol. 24, pp. 10-15. Verplanken B., Aarts H., van Knippenberg A. and Moonen A., 1998. Habit Versus Planned Behavior: A Field Experiment. In British Journal of Social Psychology, Vol. 37, pp. 111–128. 146 IADIS International Conference e-Commerce 2009 IMPROVING DIVERSITY AND RELEVANCY OF ECOMMERCE RECOMMENDER SYSTEMS THROUGH NLP TECHNIQUES Andriy Shepitsen, Noriko Tomuro College of Computing and Digital Media DePaul University, Chicago, IL U.S.A. ABSTRACT Emerging Web 2.0 technologies offer abundance of opportunities for e-commerce systems to improve the effectiveness of recommendation. For example, many e-commerce sites allow users to enter reviews together with ratings in order to obtain more feedback on their products and services. In this paper we present an approach which considers user reviews in generating personalized recommendations in e-commerce recommender systems. Our approach is novel in that the system incorporates user reviews as an additional dimension in representing the inter-relations between items and user preferences. By utilizing user reviews as the fourth dimension in addition to the traditional three dimensions of items, users and ratings, our system can generate recommendations which are more relevant to users’ interests. To analyze user reviews we utilize techniques from Natural Language Processing (NLP), a sub-field in Artificial Intelligence (AI). We extract terms/words from user reviews and analyze their parts-of-speech (POS). Then we use nouns and adjectives (only) to represent a user review, and develop a new recommendation model, which we call RecRank, that utilizes all four dimensions. We also incorporate the notion of authority – items which are frequently mentioned in other items’ reviews are considered popular and authoritative, thus should be ranked higher in the recommendations. We run several experiments on a real e-commerce data (Amazon books) and compare results with other standard recommendation approaches, namely item- and collaborative-based filtering and association rule mining algorithm. The results showed that user reviews were effective in increasing the diversity as well as relevancy of the recommended items. KEYWORDS Recommender Systems, E-Business Application, Reviews, NLP, Knowledge Discovery 1. INTRODUCTION In our highly competitive market it is very important for companies to obtain information about who their customers are, what their preferences are and how they are evaluating their existing products. The trends in customer preferences and requirements should be the main factors in navigating the company success strategy. User studies are an effective way to get information about customers preferences. However, user studies are costly and involve careful planning. With the advent of Web 2.0, many e-commerce sites encourage users to leave on-line feedback and share their opinions about the products they bought, such as Amazon (www.amazon.com) and e-bay (www.ebay.com). Users post their overall explicit ratings of the items together with reviews to explain their ratings. There are a lot of studies in e-business and psychology concerning the people’s motivation to post their reviews [1, 2]. Those studies are beyond the scope of this paper, but their general finding is that users often read reviews of others to make their own decisions about purchasing, and want to pay back and help other customers. Therefore, user reviews are a very valuable source of information for on-line e-commerce applications, such as recommender systems. Recommender Systems use information about a customer’s previous purchases, ratings and profile to predict which products he or she might be interested in buying next. There are two main technical approaches in e-commerce recommender systems: content-based and collaborative-based. In the content-based approach, recommendations are selected based on the similarity of items measured by various item features such as user ratings, author, producer, publisher, etc. In the collaborative-based approach on the other hand, 147 ISBN: 978-972-8924-89-8 © 2009 IADIS recommendations are selected based on the similarity of users and the ratings those like-minded users had entered. However, both approaches along with other hybrids suffer from the problem of “cold start”: new items cannot be found because they have only limited historical data [3, 4]. There is also another type of approach called Association Rule Mining (or the Apriori algorithm) [5]. This approach generates recommendations based on the items which often appeared together in customer transactions. Although previous research [6] has shown that the Apriori algorithm usually generates recommendations faster than other two approaches, it has difficulties with coverage: some items cannot be recommended at all. In this paper, we present a new approach which considers user reviews in addition to item similarity and user similarity in generating recommendations. We first extract terms/words from user reviews and obtain their parts-of-speech (POS). Then we use nouns and adjectives (only) as item features to enhance the contentbased approach. For instance, if terms such as “pawn”, “bishop”, ”defense” and “strategy” appeared in the review of a book, then probably it is a chess book and all other books that are reviewed using the same terms are good candidates to be associated with that book. This way, our approach can alleviate the cold start as well as the poor coverage problems. Moreover, terms in user reviews can help find other books which are related to logic and calculations, which in turn increases the diversity of the recommendations for the chess fans. Using terms in reviews can also help find like-minded users in collaborative approach. Previous psychological studies have shown that vocabulary is an expressive indicator which reveals information about the user interests, culture and personality [7]. Our hypothesis is that, if users have the same vocabulary, they probably have similar interests, of similar age, belong to similar social groups etc., thus may enjoy similar recommendations. By utilizing user reviews as the fourth dimension in addition to the traditional three dimensions of items, users and ratings, our system can generate recommendations which are more personalized to users' interests. We also introduce the notion of authoritative items. The intuition behind this notion is that, if an item was frequently mentioned by reviews of other items, it is most likely a popular item and serving as a reference point/item. Furthermore, we develop a new algorithm called RecRank which utilizes all four dimensions to better personalize the recommendation list. Finally, we report the results of running several experiments on a real e-commerce data (Amazon books). The results showed that our system outperformed other standard recommendation approaches. 2. RELATED WORK There are several works which tackled the problem of diversity and relevancy of the recommendations in ecommerce recommender systems. The problem of (poor) diversity in recommendations was first raised in Ziegler et al. [8]. They reported that the standard recommender systems failed to contribute to sales of the company as the systems only generated recommendations for the items which the users have already purchased. Then they introduced a new measure called Intra-List Similarity for recommendation generation. This metric indicates how similar the items in a given list of recommendation are to each other, thus essentially represents the degree of diversity of the recommendation list. They claimed that, although the diversity hindered the recall/precision standard metrics, it helped inform the users about new products and increased the company sales volume in a long term. MgGinty and Smyth [9] developed an algorithm called adaptive selection, which adds one item at a time in the recommendation list. They used customers feedback to determine if the next candidate item should be included in the list. They also reported that diversity helped create better recommendation lists which are more preferable for the users. To improve the relevancy of item- and collaborative-based systems, many researchers used various item features in addition to user ratings to measure the similarity between items [10]. Although those item features can improve item clustering and thematic recommendations, their information is static and narrowly scoped, and cannot take advantage of the user models, in particular the collective user efforts, expressed in user reviews. There are a few works which made an attempt to use user reviews in recommender systems. In particular, Aciar et. al [11] used consumer product reviews to improve recommendations. They first defined (manually) an ontology of product features (in the digital-camera domain), then mapped the user reviews to the ontology. They used NLP tools to analyze user reviews and categorized each sentence in a review into one of the three classes: good, bad or quality. Then the product feature mentioned in a given sentence is associated with the 148 IADIS International Conference e-Commerce 2009 sentence’s category and the node in the ontology for the feature is annotated with the category. In the recommendation generation phase, they extracted keywords from a request made by a specific user, mapped the keywords to the ontology, then generated a recommendation list customized for that user. However, they did not compare their results with other standard approaches, therefore the effectiveness of their approach is not empirically validated. 3. USING REVIEWS IN RECOMMENDATION GENERATION In this paper, we use the following notation to define relations in the dataset. A recommendation dataset D is denoted as a four-tuple: (1) D = U , I ,T , R where U is a set of users, I is a set of Items, T is a set of terms and R is a set of ratings. 3.1 Improving Similarity Measure and Association Rules in Recommendation Algorithms Recommendations in item-based recommender systems are determined based on item similarity. In particular, for a given user a recommendation list is formulated by selecting the items similar to the ones that the user had rated highly. In this work, we used the cosine measure to compute the similarity between two items, modified to include the similarity between the terms that appeared in the reviews in the following formula: cos(i, j ) = a ∑ u∈U ∑ (ru ,i * ru , j ) 2 u∈U (ru ,i ) * ∑ u∈U (ru , j ) 2 + (1 − a ) ∑ ∑ t∈T t∈T f (t , i ) * f (t , j ) f (t , i ) 2 * ∑ t∈T (2) f (t , j ) 2 where cos(i,j) is the cosine between the items i and j, ru,i is the rating score which a given user u gave to the item i, f(t,i) is a value for the term t in i that was obtained after applying the Principal Component Analysis (PCA) to the items/terms matrix (see below), and is a tuning coefficient which determines the distribution of weights between ratings and term features. For the item/term matrix, we first applied PCA to reduce the dimensionality of the matrix. There were a lot of synonyms, spelling variations and ambiguous terms in the user reviews in our dataset. Therefore, we had to take a measure to automatically find hidden connections between the items and the terms which were not explicitly stated in the original term frequency matrix. For the collaborative algorithm, we used the similar formula to compute the similarity between two users: cos(u1 , u2 ) = β ∑ i∈I ∑ i∈I (ru1 ,i * ru2 ,i ) 2 (ru1 ,i ) * ∑ i∈I (ru2 ,i ) 2 + (1 − β ) ∑ t∈T ∑ t∈T f (u1 , t ) * f (u2 , t ) 2 f (u1 , t ) * ∑ t∈T f (u2 , t ) 2 (3) Where cos(u1, u2) is the cosine between the users u1 and u2, ru1 ,i is the rating score which the user u1 gave to the item i, f(u1,t) is a value for the term t which the user u1 used that was obtained after applying PCA to the users/terms matrix, and is a tuning coefficient which determines the distribution of weights between ratings and term features. Finally for the Apriori algorithm, we modified the association rule slightly. In the standard association rule, an implication X ⇒ Ij means to add Ij in X, where X is a set of some items in I and Ij is a single item in I that is not present in X. We modified this rule to treat the terms in the reviews as transactions for finding additional lists of frequent item-sets. We also computed another frequent item set by treating the users as transactions. Then we merged the two lists to generate the candidate items to be added in the set of recommendations. Thus, our modified algorithm can alleviate the coverage problem with which the standard Apriori algorithm is known to have difficulties. 149 ISBN: 978-972-8924-89-8 © 2009 IADIS 3.2 Using Reviews in Recommendation Personalization For each of the three approaches described in the previous section, an initial list of recommended items is obtained. The next step is to re-rank them to better personalize the list to reflect the user’s interests. To this goal, we apply two methods: weighting by item popularity (or authority) and Artificial Neural Network (ANN). The idea of item popularity is inspired by the observation that items mentioned frequently in the reviews of other items are those that are well-known to the general public and serving as reference points. For instance, “...this textbook is slightly easier to read than Streetwise eCommerce.” - the customer mentioned the item in the review written to another item, thus he indirectly showed the authority of the referenced one. Therefore we can consider frequently referenced items authoritative. So if a recommendation list contained an authoritative item (but the user hasn’t purchased one yet), this item should be ranked higher than others (e.g. a “must-read” book). To compute the popularity scores of items, we represented all items in the dataset in a graph where a node/item is connected to another node/item if the first item referred to the second item in the reviews. Then we applied the Google PageRank algorithm [12] on the graph to derive the rank scores, or popularity scores, of the items. We normalized the rank scores at every iteration in the algorithm, so the scores were kept in the range between 0 and 1. We ran the algorithm iterations until the total change in the scores became less than a predefined threshold. Finally using the popularity scores obtained, we computed the final re-ranked score for each item in the recommendation list as the multiplication of its initial score by its popularity score. Output (Books’) Layer Hidden Layer2 Hidden Layer1 Intput (Review terms’) Layer Figure 1. The Artificial Neural Network Topology used for re-ranking recommended items Another method we used to re-rank the initial list of recommendations is a multi-layered ANN. We first constructed a network with two hidden layers, where the nodes in the input layer are the terms that appeared in all user reviews, and the nodes the output layer are the items (or books in our Amazon dataset). The schematic picture of the network is shown in Figure 1. We chose two hidden layers (instead of one) because the numbers of input and output nodes were quite large for our dataset (4,864 input and 15,930 output nodes), thus requiring a network which could model complex interactions between the input and output. As for the numbers of hidden nodes (43 and 181), we approximated by running PCA on user/item and item/term ratings matrices and observed the number of principal components which covered a large portion of the variability in the dataset. Then we trained the network with all items in the data using the ANN’s backpropagation algorithm. We continued the algorithm iterations for a predefined number of iterations (rather than until convergence) in order to avoid overfitting. The trained network is essentially a classifier which maps a set of terms used in the reviews of an item to the item itself. Finally we presented each item in the recommendation list to the network’s input layer and obtained the value of the output node which corresponded to the item. Then we used that value (between 0 and 1, produced by the sigmoid/logistic function applied at the output node) to multiply the initial score of the item and obtained the final re-ranked score of the item. 3.3 The RecRank Algorithm In addition to item authority and ANN, we also developed a new model for generating personalized recommendations which incorporates user reviews. In this work, there are four interconnected factors which influence user preferences in recommender systems: users, rating scores, items and review terms. We represent those information in a large matrix M of size n n where n=|U|+|I|+|R|+|T|. In other words, M is a 150 IADIS International Conference e-Commerce 2009 heterogeneous (and square) matrix where, for every user/rating/item/term, its associations with all other users/ratings/items/terms are recorded. The matrix M is symmetric, except for the sub-matrix which indicates item-to-item associations. In this sub-matrix, each entry is the number of references made in one item’s reviews to another item. Notice also that M represents the associations between rating scores and review terms quite conveniently. For example, in a row ’rating=1’, the entry for the column ’term=bad’ indicates the number of times the word “bad” appeared in all of the user reviews which gave the rating score of 1. In addition to the matrix M, we also set up another vector, which we call a personalization vector, to → personalize recommendations for a specific user. This vector p , is of size n (=|U|+|I|+|R|+|T|), and the values are binary – 1 in the slot of the user himself/herself and the slots of the items for which he/she rated highly (above his/her average rating) as well as the slots of the terms which he/she used in the reviews frequently, → and all other slots have zeros. Then by using M and p of a given user, we wish to find weights on the items – then those weights will be used to re-rank the items selected in the initial recommendation list (see below). To that goal, we defined the following formula and obtained the weights through an iterative process. → → → → (4) w1 ← α * d w1 + β * w1 * M + (1 − d )γ * p where w1 is the weighting vector of size n (=|U|+|I|+|R|+|T|), initialized with a random numbers between 0 → and 1, d is the damping factor (which helps avoid the “trap of local maximum” during iteration), p , is the personalization vector for the given user, and , , are tuning coefficients which distribute the importance of three factors influencing the weighting vector. The values of , , were determined during the → preliminary run with the training dataset. Figure 2 shows an example of M, a weight vector (W) and p . Name Rating Item Review Terms John 5 book 1 good math book John 1 book 2 bad math book Name Rating Item Mary 5 book 3 good philology book Mary 1 book 4 bad philology book book 1 book 3 book 2 book 4 Review Terms Matrix M John John Mary 1 Mary 5 book 1 book 2 2 2 5 1 1 2 2 2 1 1 2 1 2 2 3 4 book 3 5 1 book 4 good bad book math philology 2 2 1 1 2 1 1 1 1 1 1 1 1 1 3 1 4 1 5 2 book 1 5 book 2 1 2 1 1 1 1 1 book 3 5 book 4 1 2 1 1 1 1 1 1 1 1 1 1 good 2 1 1 bad 2 book 2 2 1 1 math 1 1 1 1 philology 1 1 1 16 dimensions 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 16 dimensions { { W<weighted vector>[0.234234234,0.23443...... , 0.85763522] 1 1 1 P<John’s personalization vector>[1, 0, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0 ] p value = 50% (we include 50% of John's most frequent terms) Figure 2. Matrix M, a Weight Vector and a Personalization Vector The RecRank algorithm is listed in Algorithm1. We ran this algorithm twice, with and without using the preference vector. The final re-ranking weights for the recommended items are obtained by the formula 151 ISBN: 978-972-8924-89-8 © 2009 IADIS → → → → ( w p − w) , where w p and w are the weighting vectors derived by the algorithm with and without using the personalization vector respectively. Such approach helps to cancel popular items in D which are not personalized for the target user, similar to FolkRank algorithm [13]. Input: Set of items i ∋ I, users u ∋ U, terms t ∋ T, p the personalization vector over <U,I,R,T>; simthr - a threshhold which specifies the smallest in the cosine similarity between iterations; n - maximum number of iterations Output: Weighting vector w1 over <U,I,R,T> Initialize w1 with random numbers between 0 and 1. → → w0 = w1 while (n>0) do → → → → w1 = α * d w1 + β * w1 * M − (1 − γ ) * p → → → Normalize( w1 ) ; if (Cosine( w0 , w1 ) > simthr ) then BREAK; → → w0 = w1 ; n = n - 1; End Algorithm 1. The RecRank Algorithm 4. EXPERIMENTAL EVALUATION We evaluated our approaches on the Amazon Book recommendation dataset (www.amazon.com/books). A Web crawler was used to extract data in February of 2009. The data contained 2,987 users and 15,930 books. For every user we extracted a full profile consisting of ratings and reviews he/she has posted so far. As for user reviews, we used a NLP tool called Stanford Log-linear Part-Of-Speech Tagger (http://nlp.stanford.edu/software/tagger.shtml) to identify the part of speech for each word/term, and took only nouns and adjectives and discarded all other words. We focused on nouns and adjectives because we thought those are the most semantically meaningful features. Then we we applied the Porter Stemmer algorithm [14] to further reduce the number of terms. In all, there were a total of 4,864 terms in the dataset. For evaluation metrics we used F-Score ⎛⎜ 2 * precision * recall ⎞⎟ and Intra-list Similarity (ILS)[8]. F⎜ precision + recall ⎟ ⎝ ⎠ Score is defined by precision and recall, thus measures the relevancy of the recommendations, while ILS indicates the diversity of the recommendations in the recommended list, as described previously in section Erro! A origem da referência não foi encontrada.. By plotting F-Scores against ILS, we can observe the trade-off between the relevancy and the diversity. So with this setup, results which achieved high F-Scores at high ILS values are more desirable. ILS is computed as: 2 * ∑k∈l ∑e∈l ( sim(ik , ie )) (5) ILS (l ) = 1 − n( n − 1) where l is a recommendation list, ik, ie, re items in l and n is a size of l, as per [8]. Values of both F-score and ILS are in the range between 0 and 1 where 1 is the highest (i.e., the most relevant for F-Score and the most diverse for ILS). Note that in the evaluations, we considered only the items which the user rated highly (i.e., higher than his/her average rating), as mentioned in the previous section with regard to the personalization vector. We hypothesize that we should recommend only items which the user would like. Also note that we used 5-fold cross validation for all experiments reported below. 152 IADIS International Conference e-Commerce 2009 0,6 Basic AR AR with adjectives and nouns AR with nouns 0,5 F-score 0,4 Basic CF CF with adjectives and nouns CF with nouns 0,3 Basic IBF IBF with adjectives and nouns IBF with nouns 0,2 0,1 0 0,1 0 0,2 0,3 0,4 ILS Figure 3. Relevancy and Diversity using Review Terms in Association Rules (AR), Collaborative Filtering (CF) and Item-based Collaborated Filtering (IBF) We first evaluated the contribution of review terms in the three standard recommendation approaches (Association Rules (AR), Collaborative Filtering (CF) and Item-based Collaborated Filtering (IBF)). Figure 3 shows the F-Score vs. ILS curves. As you can see, for all approaches the inclusion of the review terms, especially nouns (only), helped improve the results (i.e., relevancy and diversity of the recommendations) over the basic versions. 0,6 0,4 Basic CF CF re-ranked by ANN trained with adjectives and nouns CF re-ranked by Popularity 0,3 CF re-ranked by ANN trained with adjectives F-score 0,5 Basic IBF IBF re-ranked by ANN trained with adjectives and nouns IBF re-ranked by Popularity 0,2 0,1 0 0,1 0 0,2 0,3 0,4 IBF re-ranked by ANN trained with adjectives ILS Figure 4. Re-ranking Recommendation Lists Using Popularity and ANN Next we evaluated our recommendation re-ranking methods, as they are applied in CF and IBF. Figure 4. shows the results. There you can see that popularity (or authority) helped improve the result over the basic versions, especially for IBF. As for ANN, it seems the best improvement was achieved when the network was trained with adjectives (only). Comparing with results from the previous evaluation above, where adjectives didn’t contribute significantly in finding ’good’ initial recommendations, they seem to help in reranking. It’s probably because a “good” and “perfect” anthropology book is not an excellent candidate for recommendation to math lovers who reviewed math books using terms such as “good’’ and “excellent’’. But once an initial recommendation list is formed with books related to anthropology, adjectives such as “awesome” and “excellent” may be effective in re-ranking the initial list. 0,6 Basic CF 0,5 Basic IBF RecRank with adjectives RecRank with nouns and adjectives RecRank with nouns F-score 0,4 0,3 0,2 0,1 0 0 0,1 0,2 0,3 0,4 ILS Figure 5. Formulating Recommendation Lists Using RecRank Algorithm Finally we evaluated the RecRank algorithm. Figure5 shows the results by RecRank with nouns and/or adjectives. There you see that RecRank was only relatively effective as compared to CF and IBF – although 153 ISBN: 978-972-8924-89-8 © 2009 IADIS RecRank with nouns significantly outperformed CF, it was comparable to IBF, which shows its great potential. 5. CONCLUSION In this paper we presented the results of incorporating user reviews in e-commerce recommendation systems. User reviews are free-form texts, and we used NLP techniques to analyze them, in particular to obtain the parts-of-speech of the terms. Our experimental results showed that reviews are an important source of information about the user preferences and interconnections among items, and including reviews in the recommendation generation process does improve the relevance and diversity of the recommended items. As for recommendation re-ranking methods, popularity and ANN can also improve the quality of an already formulated recommendation list. We also proposed a new RecRank algorithm which covers all factors and their interactions and interconnections. Our future work will focus on the further improvement of the RecRank algorithm. In particular, we are interested in using other NLP techniques (in addition to POS tagging) to find terms which will influence the user preferences and item interconnections the most. We believe that, if there is a possibility of further improving the recommendations’ relevancy and diversity through reviews, it would be possible in the proposed RecRank algorithm. REFERENCES [1] P. Tallon, K. Kraemer, V. Gurbaxani, Executives’perceptions of the business value of information technology: a process-oriented approach, Journal of Management Information Systems 16 (4) (2000) 145–174. [2] M. Von Zedtwitz, Organizational learning through postproject reviews in RD, D Management 32 (3) (2002) 255–268. [3] A. Schein, A. Popescul, L. Ungar, D. Pennock, Methods and metrics for cold-start recommendations, in: Proceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval, ACM New York, NY, USA, 2002, pp. 253–260. [4] X. N. Lam, T. Vu, T. D. Le, A. D. Duong, Addressing cold-start problem in recommendation systems, in: ICUIMC ’08: Proceedings of the 2nd international conference on Ubiquitous information management and communication, ACM, New York, NY, USA, 2008, pp. 208–211. [5] R. Agrawal, R. Srikant, I. Center, C. San Jose, Mining sequential patterns, in: Data Engineering, 1995. Proceedings of the Eleventh International Conference on, 1995, pp. 3–14. [6] Y. Cho, J. Kim, S. Kim, A personalized recommender system based on web usage mining and decision tree induction, Expert Systems with Applications 23 (3) (2002) 329–342. [7] I. Nation, D. Nation, Teaching and learning vocabulary, Heinle & Heinle, 1990. [8] C. Ziegler, S. McNee, J. Konstan, G. Lausen, Improving recommendation lists through topic diversification, in: Proceedings of the 14th international conference on World Wide Web, ACM New York, NY, USA, 2005, pp. 22–32. [9] L. McGinty, B. Smyth, On the role of diversity in conversational recommender systems, Lecture notes in computer science (2003) 276–290. [10] J. Wang, A. De Vries, M. Reinders, Unifying user-based and item-based collaborative filtering approaches by similarity fusion, in: Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval, ACM New York, NY, USA, 2006, pp. 501–508. [11] S. Aciar, D. Zhang, S. Simoff, J. Debenham, Informed recommender: Basing recommendations on consumer product reviews, Intelligent Systems, IEEE 22 (3) (2007) 39–47. [12] S. Brin, L. Page, The anatomy of a large-scale hypertextual Web search engine, Computer networks and ISDN systems 30 (1-7) (1998) 107–117. [13] A. Hotho, R. Jaschke, C. Schmitz, G. Stumme, Folkrank: A Ranking Algorithm for Folksonomies, in: Proceedings of the FGIR, 2006. [14] C. Van Rijsbergen, S. Robertson, M. Porter, British Library Research and Development Dept, U. of Cambridge. Computer Laboratory, New models in probabilistic information retrieval, Computer Laboratory, University of Cambridge, 1980. 154 IADIS International Conference e-Commerce 2009 FACTORS AFFECTING ONLINE APPLICATION OF INSURANCE PRODUCTS AND ITS IMPLICATIONS Wee Kheng-Tan Assistant Professor, Department of Information & Electronic Commerce, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County, Taiwan Yu-Jie Tan Department of Information & Electronic Commerce, Kainan University, Kainan University, No. 1, Kainan Road, Luchu, Taoyuan County, Taiwan ABSTRACT Insurance sales agents play a very important role in the insurance industry. As the first point of contact with customers, they are instrumental in promoting insurance products of the insurance company. Popularity of electronic commerce has pushed many insurance companies to offer their customers the option of buying insurance products online. Consumers can now purchase insurance products independent of direct service employee involvement, hence creating an impact on the traditional one-to-one marketing employed by insurance sales agents. In addition, it is also known that electronic commerce is not necessarily suitable for all types of products. Given the complicated nature of insurance products, it may be unlikely for Internet to replace insurance sales agents. Furthermore, online mode introduces a new set of online-related criteria which customers need to factor into their decision when they apply online. Since the criteria considered by customers when applying online might not be of equal importance, this research used the Analytic Hierarchy Process to study the impact of the inclusion of these online-related criteria. In addition, this research used the more simple travel insurance and the much more complicated individual life insurance for comparison purpose. This research observed that reasonable premium remained the most important criteria and was applicable to the above two types of insurance. Online-related criteria featured more prominently when customers applied for travel insurance online than for individual life insurance. Insurance sales agents were slowed to appreciate the importance of online-related criteria when customers applied for travel insurance online. On the other hand, the impacts of online-related criteria when customers applied life insurance were less important, resulting in greater agreement between customers and insurance sales agents. KEYWORDS Travel Insurance, Life Insurance, Online Application. 1. INTRODUCTION Risk may be defined as uncertainty concerning the occurrence of a loss (Rejda, 2008). Insurance involves the transfer of risks to insurers who agree to indemnify insureds for such loss in exchange for a premium. Insurance is popular among the Taiwanese. In term of global ranking, Taiwan was ranked 14th in premium income, 1st in insurance penetration (ratio of insurance premium to GDP) and 19th in insurance density (average insurance expense per capita) in 2007 (Swiss Re, 2008). Based on the number of business licenses issued, Taiwan has 61 insurance institutions in 2008 (Taiwan Insurance Institute, 2009) and is an indication of the liberalization of the insurance industry and the tough competition faced by industry players. As a form of service, insurance sales agents play a very important role. Traditionally, insurance sales agents are often the first point of contact for those who intend to buy insurance policies. Through introducing, explaining, processing and notifying the approval of insurance policies, insurance sales agents help customers to have a better understand of the insurance product and generate income for the insurance company. Furthermore, insurance sales agents can also help to build enduring customer relationship. In return, insurance company offers them competitive commission to spur them on. As a result, insurance industry often gives outsiders the impression of being a high-contact service. 155 ISBN: 978-972-8924-89-8 © 2009 IADIS Popularity of electronic commerce has pushed many insurance companies to offer customers the option of buying insurance products online. This online option can be viewed as an additional mean of service delivery and a form of self-service technology facility. (Meuter et al., 2000). Customers can now purchase insurance products independent of direct service employee involvement. This move will definitely have an impact on the traditional one-to-one relationship-based marketing employed by insurance sales agents. From the customers’ perspective, customers also need to take into account some of the online-related issues such as user interface and security when they intend to purchase insurance product online. Online option may drive a wedge between insurance sales agents and customers. Since the former are supposed to understand the customer best, it is interesting to understand whether online application has resulted in insurance sales agents having a poorer understanding of their customers. It is also known that electronic commerce is not always suitable for all types of products; often the more complicated the product, the less it can do well in the cyberworld. Some researchers e.g. Rogers (2001) felt that given the complicated nature of insurance products, it was unlikely for Internet to replace traditional insurance sales agents. Careful observation reveals that insurance products are not uniform in complexity. As an example, travel insurance, which covers medical expenses, financial and other losses incurred while traveling, is a relatively simple product. However, individual life insurance is much more complicated. Besides the basic life insurance policy, it is common for endorsements and riders to be added to the policy. In view of the above, this research used two types of insurance products: the simple travel insurance and the much more complicated individual life insurance for investigation and comparison purpose. They could be viewed as two products at two opposite point of the spectrum in term of the need for service contact. We noticed that customer consider many criteria when deciding whether to purchase the insurance product. We added a set of online-related criteria to find out whether there were important differences in customers’ consideration when they applied online for these two types of products. As these criteria were not necessarily of equal importance, Analytical Hierarchy Process (AHP) method was used. This research also tries to find out whether insurance sales agents understood their customers well enough in the online environment. 2. LITERATURE REVIEW Service encounter is one of the three stages of service consumption where customer interacts with the service provider. It can range from high contact services such as hotel to low contact services such as filling the car with petrol (Curran et al., 2003). Service encounter generates the most powerful perception of service (Sweeney and Wade, 2000), affects customer’s perceived service quality (Meuter et al., 2000) which may contribute to satisfaction (Churchill and Carol, 1982) and re-purchase intention (Taylor and Baker, 1994). As mentioned earlier, insurance industry is often perceived as a relatively high contact service with insurance sales agents playing a key role. However, we also note that different insurance products needs different level of contact service. Nevertheless, it is important for insurance company to be concerned about the types of service they offered and the associated quality. Service offered could include the insurance sales agents themselves, call centers, overseas rescue and emergency services. Customer loyalty is viewed as the strength of the relationship between an individual’s relative attitude and repeat patronage (Dick and Kunal, 1994). Loyalty helps guarantee future earnings of an organization (Sharp and Sharp, 1997). Besides repeat business, insurance company also hopes that existing customers will also purchase other products offered by the same company. Corporate reputation is a result of the past actions of a firm and serves to communicate to its target groups information regarding the quality of its products or services in comparison with those of its competitors (Yoon et al., 1993). It is also built through its credible actions (Nguyen and Leblanc, 2001) and a symbol that it is worthy to believe (Gupta and Cooper, 1992). Facing asymmetric information or product attributes which are different to evaluate prior to purchase, corporate reputation serves as a useful signal for the consumer. As a financial instrument and sometimes with a long effective duration and maturity period, the reputation of insurance company is thus important. The online option offered by insurance company usually includes insurance product information, online application and approval as well as allowing customer to make the necessary payment. Hence when deciding whether to buy a policy online, besides considering price, coverage, compensation when the undesirable happened, services etc., customer will also need take into account some of the online-related issues such as 156 IADIS International Conference e-Commerce 2009 user interface, online approval, secured handling of personal data as well as able to make payment online. Related to the user interface is the concept of ease of use. An important measure of the Technology Acceptance Model (TAM), ease of use is defined as ‘the degree to which a person believes that using a particular system would be free from effort’ (Davis, 1989). TAM is widely used to explore why users adopt various systems and applications. 3. RESEARCH METHOD The Analytic Hierarchy Process (AHP) was first proposed by Thomas L. Saaty (Saaty 1977) and its main characteristic is based on pairwise comparison judgments. It is often used to analyze complex multi-criteria decision-making problem and applied to a wide range of problems and applications. The problem investigated is set up as a sequential step-up hierarchy structure with different measures being integrated into a single overall score for ranking decision alternatives. When using AHP, the problem’s step-up hierarchy structure need to be developed. A hierarchy is a specific type of system, based on the assumption that the entities can be grouped into disjoint sets, with the entities of one group influencing those of other groups (Ananda and Herath, 2002). Pairwise comparison matrix forms an important component of AHP. Two criteria are compared using a nine-point scale, where ‘1’ denotes ‘equal’ importance, ‘3’ represents ‘slightly’ importance, ‘5’ indicates ‘clearly’ superiority, ‘7’ is ‘very’ important and ‘9’ denotes ‘extremely’ important, with the even numbers being used to indicate intermediate values if necessary. If there are n criteria to consider, n(n-1)/2 pairwise comparisons would need to be made. Afterwhich, reciprocal n x n matrix are constructed and weightings are then obtained. The consistency of the pairwise comparison matrix needs to be verified and the following two indexes can be used: Consistency Index (C.I.) and Consistency Ratio (C.R.). They are defined as follows: C.I. = ( λmax – n ) / ( n – 1) (1) C.R. = C.I. / R.I. (2) where λmax is the largest eigenvalue and R.I. (Random Index) is as shown in Table 1. For consistency, C.I and C.R. should be less than 0.1 for the AHP analysis to be acceptable. Table 1. Random Index (R.I.) N R.I. 1 0 2 0 3 0.58 4 0.90 5 1.12 6 1.24 7 1.32 8 1.41 9 1.45 10 1.49 4. RESEARCH FRAMEWORK, RESULT AND DISCUSSION The relevant literature was used as a starting point to identify the key criteria suitable for this research. Discussions with insurance sales agents and customers were then made to refine the criteria. A tri-tiered hierarchical structure as shown in Figure 1 was obtained. The second tier comprises three aspects which are firstly, operation of insurance company, secondly, key elements of insurance policy and thirdly, online application and processing. The third tier comprises 11 criteria. As the AHP questionnaire was not easy to fill up, face-to-face survey was conducted to increase the accuracy of the returns. The survey lasted for 2 months starting from December 5, 2008. A total of 44 valid returns were received from insurance sales agents and 102 valid returns were obtained from customers. Fiftyone valid returns were obtained for online application of travel insurance and another 51 valid returns for online application of individual life insurance. Eighty percent of the insurance agents were male, 65% of them were in the age group of 19-25, and 64% received college and university education. In addition, 86% of them were full-time insurance sales agents. For the customers, 63% of them were male and 92% were in the main age group of 19-35 year old. In addition, 51% of them received college and university education. 157 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 1. The Hierarchical Structure 4.1 Online Travel Insurance As observed from Table 2, customers considered online application and processing (0.386) to be the most important aspect, followed by key elements of insurance policy (0.325). They are least concerned about operation of insurance company (0.289). Table 2. Weighting of Online Travel Insurance Application (Customer) Aspects and Criteria A. Operation of Insurance Company A1. Reputation of Insurance Company A2. Customer Service Offered by Insurance Company A3. Existing Customer of Insurance Company A4. Overseas Service Offered by Insurance Company B. Key Elements of Insurance Policy B1. Reasonable Premium B2. Flexible Range of Risks Covered by the Insurance Policy B3. Reasonable Compensation C. Online Application and Processing C1. Easy to Use Interface C2. Simple Approval Process C3. Many Payment Options C4. Security Note: Consistency Index (C.I.) and Consistency Ratio (C.R.) < 0.1 Aspect 0.289 (3) Weighting Criteria Overall 0.291 (2) 0.338 (1) 0.099 (4) 0.272 (3) 0.084 (7) 0.097 (6) 0.029 (11) 0.079 (8) 0.461 (1) 0.193 (3) 0.346 (2) 0.150 (1) 0.063 (9) 0.112 (4) 0.303 (2) 0.273 (3) 0.110 (4) 0.314 (1) 0.117 (3) 0.105 (5) 0.043 (10) 0.121 (2) 0.325 (2) 0.386 (1) However, as shown in Table 3, insurance sales agents were of the opinion that their customers would be most concerned with key elements of insurance policy (0.379), followed by operation of insurance company (0.344). The aspect which customers were most concerned with, online application and processing, was ranked by the insurance sales agent as the least important and with a weighting of 0.277. This outcome reflected a fundamental difference between the understanding of customers and insurance sales agents. 158 IADIS International Conference e-Commerce 2009 Table 3: Weighting of Online Travel Insurance Application (Insurance Sales Agent) Aspects and Criteria A. Operation of Insurance Company A1. Reputation of Insurance Company A2. Customer Service Offered by Insurance Company A3. Existing Customer of Insurance Company A4. Overseas Service Offered by Insurance Company B. Key Elements of Insurance Policy B1. Reasonable Premium B2. Flexible Range of Risks Covered by the Insurance Policy B3. Reasonable Compensation C. Online Application and Processing C1. Easy to Use Interface C2. Simple Approval Process C3. Many Payment Options C4. Security Note: Consistency Index (C.I.) and Consistency Ratio (C.R.) < 0.1 Aspect 0.344 (2) Weighting Criteria Overall 0.330 (2) 0.352 (1) 0.115 (4) 0.203 (3) 0.113 (5) 0.121 (3) 0.039 (10) 0.070 (7) 0.465 (1) 0.200 (3) 0.335 (2) 0.176 (1) 0.076 (6) 0.127 (2) 0.247 (2) 0.188 (3) 0.133 (4) 0.432 (1) 0.069 (8) 0.052 (9) 0.037 (11) 0.120 (4) 0.379 (1) 0.277 (3) As for the criteria under operation of insurance company, rankings given by customers and insurance sales agents were identical. Customer service offered by insurance company was the most important, followed by reputation of insurance company, overseas service offered by insurance company and lastly existing customer of insurance company. This result showed that loyalty of existing customers was not important. For key elements of insurance policy, criteria which insurance sales agents felt customers would be most concerned with and what customers really felt, i.e. reasonable premium, was identical. The rankings given by both parties for reasonable compensation and flexible range of risks covered by the insurance policy were identical and ranked second and third respectively. For criteria under online application and processing, rankings given by customers and insurance sales agents were identical but the weighting revealed interesting messages. Security emerged as the first. However, the weighting given by insurance sales agent for security (0.432) was much higher than the weighting given by customers for security (0.314). Easy to use was ranked second by both parties, however, it was valued more by customers (0.303) than insurance sales agents (0.247). Through the overall ranking analysis, it was observed that reasonable premium was top in the mind of both customers and insurance sales agents. Security and easy to use interface were also important to customers. However, insurance sales agents generally under-estimated the importance of online-related criteria to the customers. Security, easy to use interface and simple approval process were ranked second, third and fifth in importance by the customers. However, insurance sales agents thought that their customers would ranked them only as fourth, eight and ninth respectively. 4.2 Online Individual Life Insurance As shown in Table 4, customers considered operation of insurance company (0.417) to be the most important, then key elements of insurance policy (0.361). They were least concerned about online application and processing. Insurance sales agents (Table 5) also felt that customers would rank online application and processing (0.257) last. However, insurance sales agents felt that their customers would instead be most concerned about elements of insurance policy (0.444), followed by operation of insurance company (0.299). For the criteria of operation of insurance company, customers valued reputation of insurance company most, followed by the customer service offered by insurance company. However, insurance sales agents thought that customers would value customer service most, followed by reputation. Existing customer of insurance company was ranked last by both parties, reflecting the consensus that it was not that important. For key elements of insurance policy, the criteria which insurance sales agents felt customers would be most concerned with and what customers really felt was identical, i.e. reasonable premium. However, insurance sales agents under-estimated the importance of reasonable premium, customers gave a high weighting of 0.505 while agents gave a weighting of 0.385. Rankings given by both parties for reasonable compensation and flexible range of risks covered by the insurance policy were identical, i.e. ranked second and third respectively. For online application and processing, customers considered security as the most important, 159 ISBN: 978-972-8924-89-8 © 2009 IADIS followed by easy to use interface and simple approval process. However, insurance sales agents thought that customers would value simple approval process most, followed by security and easy to use interface. Table 4. Weighting of Online Individual Life Insurance Application (Customer) Aspects and Criteria A. Operation of Insurance Company A1. Reputation of Insurance Company A2. Customer Service Offered by Insurance Company A3. Existing Customer of Insurance Company A4. Overseas Service Offered by Insurance Company B. Key Elements of Insurance Policy B1. Reasonable Premium B2. Flexible Range of Risks Covered by the Insurance Policy B3. Reasonable Compensation C. Online Application and Processing C1. Easy to Use Interface C2. Simple Approval Process C3. Many Payment Options C4. Security Note: Consistency Index (C.I.) and Consistency Ratio (C.R.) < 0.1 Aspect 0.417 (1) Weighting Criteria Overall 0.339 (1) 0.318 (2) 0.109 (4) 0.234 (3) 0.141 (2) 0.133 (3) 0.045 (10) 0.098 (5) 0.505 (1) 0.168 (3) 0.327 (2) 0.182 (1) 0.061 (8) 0.118 (4) 0.297 (2) 0.206 (3) 0.121 (4) 0.376 (1) 0.066 (7) 0.046 (9) 0.027 (11) 0.083 (6) 0.361 (2) 0.222 (3) Table 5. Weighting of Online Individual Life Insurance Application (Insurance Sales Agent) Aspects and Criteria A. Operation of Insurance Company A1. Reputation of Insurance Company A2. Customer Service Offered by Insurance Company A3. Existing Customer of Insurance Company A4. Overseas Service Offered by Insurance Company B. Key Elements of Insurance Policy B1. Reasonable Premium B2. Flexible Range of Risks Covered by the Insurance Policy B3. Reasonable Compensation C. Online Application and Processing C1. Easy to Use Interface C2. Simple Approval Process C3. Many Payment Options C4. Security Note: Consistency Index (C.I.) and Consistency Ratio (C.R.) < 0.1 Aspect 0.299 (2) Weighting Criteria Overall 0.289 (2) 0.331 (1) 0.153 (4) 0.227 (3) 0.086 (5) 0.099 (4) 0.046 (10) 0.068 (8) 0.385 (1) 0.272 (3) 0.343 (2) 0.171 (1) 0.121 (3) 0.152 (2) 0.215 (3) 0.335 (1) 0.168 (4) 0.282 (2) 0.056 (9) 0.086 (5) 0.043 (11) 0.072 (7) 0.444 (1) 0.257 (3) The overall ranking analysis revealed that reasonable premium was top for both customers and insurance sales agents with customers giving this criteria a higher overall weighting (0.182) than insurance sales agents (0.171). Reputation of insurance company came in second (0.141) which insurance sales agents underestimated (0.086). Among the criteria which insurance sales agents over-estimated their importance, flexible range of risks covered by the insurance policy and simple approval process were the most over-estimated. 4.3 Comparison of Online Travel Insurance and Individual Life Insurance When the three aspects of the two insurance products are compared, customers considered online application and processing (0.386) as the most important for online travel insurance but operation of insurance company (0.417) was the most important for online individual life insurance. However, the insurance sales agents thought that, regardless of the type of insurance, customers would rank key elements of insurance policy as the most important, followed by operation of insurance company and lastly online application and processing. This observation suggested that insurance sales agents might have the same mindset when they faced different types of insurance even though this was not so for the customers themselves. For the operation of insurance company, customers felt that reputation of insurance company and customer service offered by insurance company were the top two criteria even though customers placed 160 IADIS International Conference e-Commerce 2009 higher emphasis on reputation when purchasing online individual life insurance. As for the insurance sales agents, they thought that customers would, again regardless of the type of insurance, placed top emphasis on customer service offered by insurance company followed by reputation of insurance company. As for key elements of insurance policy, all parties considered reasonable premium as the most important, followed by reasonable compensation. Regardless of the type of insurance, customers valued security and easy to use interface most. For all the parties, many options for payment was not an important criteria. 4.4 Overall Discussion Viewing the above in totality, one can draw the conclusion that the nature of the insurance product affects customers’ consideration. Even though travel insurance and individual life insurance are insurance products, the former is much simpler, structured, cheaper and of shorter maturity period. However, the latter is complicated, less structured, expensive and with maturity period stretching maybe over many years. Hence, when purchasing individual life insurance, customers will definitely be more cautious, spend more time to consider and prefer the guidance and assistance of insurance sales agents. It is a product that needs higher contact service than travel insurance. The ranking in importance given by the customers for both types of insurance products reflect the effects of such product’s nature. Reasonable premium is top on the list of importance, hence pricing constitutes a very critical part of customers’ decision. The overall reasonable premium weighting of individual life insurance (0.182) is higher than travel insurance (0.150), reflecting the effect of the more expensive individual life insurance. Even though customers are very concerned of the amount they are paying, they are less concerned with what they will get in return (i.e. compensation). For both types of insurance, customers have only ranked reasonable compensation fourth in overall importance, which is lower than the second ranking given by insurance sales agents. Such outcome is obtained because compensation is received only upon some undesirable happenings and of course, many would prefer such events never to happen to them. Furthermore, many customers buy insurance policies more to get the value derived from the ‘peace of mind’ than the hope for high actual claim. The similarity of these two types of insurance products ends at reasonable premium. For the online travel insurance, the immediate set of criteria of high importance are mostly related to online application and processing: security (ranking: 2), easy to use interface (ranking: 3) and simple approval process (ranking: 5), follows by the next set of criteria which are more related to the insurance company and service offered: customer service offered by insurance company (ranking: 6) and overseas service offered by insurance company (ranking: 8). Given the simple and structured feature of travel insurance, there is no real need for active involvement of insurance sales agents, thereby making online application a truly viable option. Furthermore, online application offers convenience for buyers anywhere and anytime which is often needed because customers tend to buy travel insurance at the last moment. Hence, criteria related to online application and processing are predominant. On the other hand, for the case of individual life insurance, reputation of insurance company is ranked second, then follows by a set of criteria related to services offered: customer service (ranking: 3) and overseas services (ranking: 5) offered by insurance company. Online-related criteria come later: security (ranking: 6), easy to use interface (ranking: 7) and simple approval process (ranking: 9). Customers are very concerned about the reputation of insurance company because life insurance is more expensive with long maturity period, hence the need for a reputable insurance company to better protect their interest. This is even more important now because of the ongoing financial crisis. Due to the fact that individual life insurance is much more complicated, often with riders attached to a standard life insurance policy, the need for personalized service and professional advice offered by the insurance sales agent is obvious here. Furthermore, the complexity of this product makes full automation and self-service impossible. Application may be made online but insurance sales agent often need to pick up the lead and do the necessary contact with the applicant, follow-up and processing. It thus causes online-related criteria to be less critical than online travel insurance. Loyalty and many options for payment are at the tail-end of the list of overall importance. With premium as the customer’s key concern and faces with many competitive options offered by other insurance companies, customer’s loyalty to the insurance company is low. It is also common for customers to diversify their risk and buy policies from more than one insurance company. Payment is slightly important for online travel insurance application (ranked 10) because of the need to make online payment. 161 ISBN: 978-972-8924-89-8 © 2009 IADIS The main difference in viewpoint of customers and insurance sales agents is due to the online travel insurance and specifically contribute by the online-related criteria. Insurance sales agents are slowed to appreciate the suitability of e-commerce for travel insurance and hence the importance of online-related criteria when customers applied for travel insurance online. Insurance company will need to establish other feedback mechanism to receive customers’ feedback on the online options, especially for insurance products which are easier to automate online instead of relying on insurance sales agents for feedback and suggestions. 5. CONCLUSION This research showed that the importance of criteria differed according to the complexity of the insurance product. Through the AHP method, this research observed that reasonable premium was the most important criteria. Online-related criteria featured more prominently for customers when they applied for the much simpler travel insurance online than for the more complicated individual life insurance. Furthermore, insurance sales agents were slowed to appreciate the importance of online-related criteria when customers applied for travel insurance online. On the other hand, the impacts of online-related criteria on customers when they applied for individual life insurance were much less important, resulting in greater agreement between customers and insurance sales agents. REFERENCES Ananda J. and Herath G., 2002. Assessment of Wilderness Quality Using the Analytic Hierarchy Process. In Tourism Economics, Vol. 8, No. 2, pp. 162–182. Churchill G. A. Jr. and Carol S., 1982. An Investigation into the Determinants of Customer Satisfaction. In Journal of Marketing Research, Vol. 19(November), pp. 491-504. Curran J. M., Meuter, M. L. and Suprenant C. F., 2003. Intentions to Use Self-service Technologies: A Confluence of Multiple Attitudes. In Journal of Service Research, Vol. 5, No. 3, pp. 209–224. Davis F. D., 1989. Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. In MIS Quarterly, Vol. 13, No. 3, pp. 319-340. Dick A. S. and Kunal B., 1994. Customer Loyalty: Toward an Integrated Conceptual Framework. In Journal of the Academy of Marketing Science, Vol. 22, No. 2, pp. 99-113. Gupta S. and Cooper L. G., 1992. The Discounting of Discounts and Promotion Thresholds. In Journal of Consumer Research, Vol. 19(December), pp. 401–411. Meuter M. L., Ostrom A. L., Roundtree R. I. and Bitner M .J., 2000. Self-service Technologies: Understanding Customer Satisfaction with Technology-based Service Encounters. In Journal of Marketing, Vol. 64(July), pp. 50-64. Nguyen N. and Leblanc G., 2001. Corporate Image and Corporate Reputation in Customers' Retention Decisions in Services. In Journal of Retailing and Consumer Services, Vol. 8, pp. 227-236. Rejda G., 2008. Principles of Risk Management and Insurance 10th Edition, Addison Wesley, Boston, USA. Rogers J., 2001. Six Keys to B2C E-commerce Success. In Insurance & Technology, July, pp. 49-55. Saaty T. L., 1977. A Scaling Method for Priorities in Hierarchical Structures. In Journal of Mathematical Psychology, Vol. 15, pp. 234-281. Sharp B. and Sharp A., 1997. Loyalty Programs and Their Impact on Repeat-purchase Loyalty Patterns. In International Journal of Research in Marketing, Vol. 14, No. 5, pp. 473-486. Sweeney J. C. and Lapp W., 2000. High Quality and Low Quality Internet Service Encounters. ANZMAC 2000 Visionary Marketing for the 21st Century: Facing the Challenge, pp. 1229-1233. Swiss Re, 2008. World insurance in 2007: Emerging Markets Leading the Way. Sigma No. 3/2008. Taiwan Insurance Institute, 2009. The Important Indexes of Insurance Industry. Taiwan, February. Taylor S. A. and Baker T. L., 1994. Assessment of the Relationship Between Service Quality and Customer Satisfaction in the Formation of Consumers’ Purchase Intentions. In Journal of Retailing, Vol. 70, No. 2, pp. 163–178. Yoon K. Y. and Im K. S., 2005. An Evaluating System for IT Outsourcing Customer Satisfaction Using the Analytic Hierarchy Process. In Journal of Global Information Management, Vol. 13, No. 4, pp. 55–78. Yoon E., Guffey H. G. and Kijewski V., 1993. The Effects of Information and Company Reputation on Intentions to Buy a Business Service. In Journal of Business Research, Vol. 27, pp. 215-28. 162 IADIS International Conference e-Commerce 2009 THE CONFLICTING ROLE OF ONLINE SWITCHING COSTS: THE MAIN AND INTERACTION EFFECTS ON CUSTOMER RETENTION Ana Isabel Torres, Francisco Vitorino Martins Faculty of Economics - University of Porto Rua Dr. Roberto Frias, 4200-464 Porto - Portugal ABSTRACT In this paper we examine prior research, which has been studying the effects of switching costs on customer retention, within online and offline markets. The empirical studies in the area reveal contrasting results: (1) switching costs are a stronger predictor of customer retention than customer satisfaction alone; (2) the main effect of switching costs moderates the relationship between satisfaction and customer retention; (3) conversely, the lack of main effects reflects the interaction effects indicating that switching costs have an even greater impact on customer retention than under satisfaction variation levels. These power asymmetries reveal a conflicting role of switching costs across industry and marketplaces. We also observe a significant bias in favor of service industries with high switching costs. Based on this meta-analytical research we propose a theoretical base to support the path structural relationship between the constructs providing a nomological framework to future research. Finally we draw some conclusions, suggest future research directions and make managerial recommendations. KEYWORDS Switching Costs, Customer Retention, Electronic Markets. 1. INTRODUCTION Online switching costs are a relevant and challenging research subject due to the intense competition in the electronic markets, high customer acquisition costs and the perceived ease with which customers can switch between online suppliers. Typically, assumptions are made to confirm switching costs as important antecedents of customer retention or loyalty, and within an online environment this relationship is of critical importance. However, this relationship is perceived at different levels resulting in a lack of significant association with repurchase intentions. It is generally accepted that customer satisfaction is thought to be one of the most important determinants of customer loyalty and this relationship can be reciprocal (Oliver, 1999, Szymanski and Henard, 2001; East et al. 2008). Accordingly, Shankar et al. (2003, one of the most downloaded articles cited by Scopus) found that loyalty and satisfaction have a reciprocal relationship such that each positively reinforces the other, and this relationship is further strengthened online, meaning that loyalty to the service provider is higher when the service is chosen online than offline. Additionally, a large body of evidence from e-commerce studies, supports that higher levels of customer satisfaction reduces perceived benefits of switching behaviour and generates higher levels of loyalty (Chiou, 2004; Anderson and Srinivasan, 2003; Devaraj et al., 2002). Although the relationship between satisfaction and loyalty can be non-linear and only after satisfaction increases above a critical level does loyalty increase (Dick & Basu, 1994). In fact, the asymmetrical relationship between satisfaction and loyalty may be an indicator of the presence of moderating variables (e.g., switching barriers, purchase involvement) or where there is a priori reason for expecting moderating effects. Consistent with these findings Oliver (1999) has suggested that satisfaction is a necessary step in loyalty formation but becomes less significant as loyalty is achieved through other mechanisms. As the primary driver for explaining customer retention, customer satisfaction has received unflagging attention in the marketing literature. While evidence accumulates that satisfaction influences repeat purchase behavior, it typically explains only a quarter of the variance in behavioral intentions (see the meta-analysis by Szymanski 163 ISBN: 978-972-8924-89-8 © 2009 IADIS and Henard, 2001 for an extent review). In addition, satisfaction in predicting future behavior has been examined, and is shown to be only a weak predictor of behavioral loyalty, especially when there were high switching costs (Balabanis et al., 2006). Moreover, the relationship between satisfaction and loyalty is now recognized as more complex than originally proposed (Oliver, 1999; Mittal and Kamakura, 2001). Customers can be satisfied, yet they might not be loyal to a retailer, due to individuals’ idiosyncrasies and market contingencies. Yet, firms appear to be stuck in a “satisfaction trap” a myopic belief that customer satisfaction and service quality are the only tools for managing customer retention (Reichheld and Schefter, 2000). The relevant interest of studying switching costs, in a managerial perspective, is the premise that these costs impede customer switching and hence improve customer retention. While a large extent of literature explores this relationship, empirical evidence is not consensual about the different roles of switching costs on customer retention and satisfaction. Actually, researchers have recently called for devoting more efforts to understanding customer behavior on online services (Yang and Peterson, 2004; Zeithaml et al., 2002). Given the dramatically reduced set up and search costs in the Internet, and the perceived ease with which customers can switch between online suppliers, what are the main drivers of customer e-loyalty? What is the role of switching costs in the satisfaction-loyalty relationship? What are the main and moderating effects of switching costs in this relationship? To answer these questions, this research attempts to examine the consistency of the strength and direction of the main effects of switching costs on customer retention and the interaction effects on the relationship between satisfaction and customer retention. Regarding the argument that perceptions of switching costs may vary due to different industry characteristics (Burnham et al., 2003; Jones et al., 2002) we examine the switching costs effects across-industry-context online and offline. We use a meta-analysis as a literature synthesis, to observe and combine results from multiple experiments to arrive at a clear understanding of some effect of interest. We collected the relevant consumer switching cost literature from the leading academic journals and databases, over the last decade. We selected the studies, which have empirically investigated switching costs as a measured variable and satisfaction or customer retention as dependent variables, using Structural Equations Model (SEM), or another adequate method, to assess and test the nomologigal validity of constructs’ relationships. We retain about ten studies that more fully explain the full range of different effects of switching costs. Helpfully, we intend to contribute to a more comprehensive understanding of the importance and role of switching costs on customer retention, clarifying more complex customer behavior. Also, giving strength to the nomological framework validity, we aim to provide a theoretical base to the structural model development. Following this, we will discuss exactly what is meant by consumer switching costs, satisfaction and loyalty. Next, we will analyze the magnitude, direction and strength of different effects of switching costs on structural relationships between satisfaction and customer loyalty and we present the final conclusions. 2. THE ROLE OF SWITCHING COSTS: THE RELATIONSHIP BETWEEN SATISFACTION AND CUSTOMER RETENTION Switching costs are defined as the one-time costs that customers associate with the process of switching from one provider to another. While switching costs must be associated with the switching process, they need not be incurred immediately upon switching. Furthermore, switching costs need not be limited to objective, "economic" costs. Switching costs may perceive impediments ranging from "search costs, transaction costs, learning costs, loyal customer discounts, customer habit, emotional costs and cognitive effort, linked to financial, social and psychological risk on the part of the buyer" (Fornell et al., 1996:10). Drawing on the economic cost concept (Besanko and Braeutigam, 2005:223) switching costs can be understood as the opportunity cost of an alternative and includes all of the explicit and implicit costs (which involve and do not involve outlays of cash, respectively) associated with that alternative. Put simply, switching costs measure the full amount of sacrifices made at the moment the decision of switching provider is taken. Burnham et al. (2003) found evidence for several distinct switching costs organized in three-order switching cost types: 164 IADIS International Conference e-Commerce 2009 “Procedural switching costs” consisting of economic risk, evaluation, learning and setup costs, primarily involving the expenditure of time and effort when an individual intends to begin a relationship with a new provider, also described by Jones et al. (2002) as pre-switching costs and uncertainty costs. “Financial switching costs” consisting of benefit-loss and financial-loss-costs involving the loss of contractual benefits for staying with an incumbent firm (e.g. discounts, points) and onetime financial outlays incurred in switching providers (e.g. deposits, initiation fees) “Relational switching costs” consisting of personal-relationship loss and brand-relationship-loss costs, this type of switching costs involves psychological or emotional discomfort due to the loss of identity and breaking of bonds when individuals switch providers. Switching barriers are similar to switching costs and are defined as the degree to which customers experience a sense of being "locked into" a relationship based on the economic, social, or psychological costs associated with leaving a particular service provider. Actually, both constructs have been used simultaneously to predict customer retention (Shin and Kim, 2007; Tsai and Huang, 2007; Tsai et al. 2006; Balabanis et al. 2006; Ranaweera and Prabhu, 2004; Jones et al. 2000). Customer loyalty is defined as a buyer’s overall attachment or deep commitment to a product, service, brand or organization (Oliver, 1999). The loyalty concept manifests itself in a variety of behaviours: repeated purchases (e.g. behavioral loyalty) and commitment (e.g. attitudinal loyalty) and both measures are important. Attitudinal loyalty includes different levels varying from repurchase intentions, recommending a service provider to other customers, repeatedly patronizing the provider, willingness to pay more and ultimate loyalty (i.e. resistance to others). Several studies use both measures as loyalty indicators. Satisfaction or overall satisfaction has often been used to measure e-business success. Theoretically, overall satisfaction can be considered an affective-based construct and is generally defined as a positive affective state resulting from a global evaluation of performance based on past purchasing and consumption experience (Oliver, 1999; Fornell et al. 1996; Lam et al. 2004; Szymanski and Henard, 2001). 2.1 The Conflicting Role of Switching Costs There is empirical evidence of switching costs directly influencing customer loyalty both on a business-toconsumer offline and online context (Burnham et al. 2003, Ranaweera and Prabhu, 2004; Jones et al. 2000; Yang and Peterson, 2004; Balabanis et al. 2006;). Moreover, it has been claimed that switching costs can explain more powerfully customer loyalty, and in an online environment this relationship is enhanced (Balabanis et al. 2006; Shankar et al. 2003). Despite this main effect, we have found across recent empirical studies, more complex and different effects in the satisfaction-loyalty relationship, under the interaction of switching costs. Research findings reporting both online and offline studies are summarized in Table 1. 2.1.1 Asymmetrical Effect between Satisfaction and Switching Costs Researchers have suggested that switching costs and satisfaction may negatively interact with each other in driving customer intentions. As switching cost rise, the influence of satisfaction on intentions to stay with a provider may decrease (and vice versa). In a leading study, using a global measure of switching cost in surveying different samples of bank and hairstylist clients, Jones et al. (2000) found evidence of this negative interaction effect. They found that the influence of core-service satisfaction on repurchase intentions decreases under conditions of high switching barriers. Contrarily to Jones and colleagues, Burnham et al. (2003) based on interviews of business consumers of a credit car firm did not find support for the negative interaction between satisfaction and switching costs. They found that satisfaction alone explains only 16% of the variance on customer retention, well within the bounds reported as typical by Szymanski and Henard´s (2001) meta-analysis of satisfaction studies. They also found that switching costs have an even stronger effect on customer retention, explaining 30% of the 165 ISBN: 978-972-8924-89-8 © 2009 IADIS Table 1. Empirical Research Findings of Switching Cost Effects Across Studies Study Antecedents / Moderator* Outcome Method / Data Community building Overall Satisfaction Switching Barriers Customization Repurchase Intentions Experience value sharing Perceived switching costs Community building Perceived Service Quality Perceived trust Switching Barriers Satisfaction Relational Orientation* Satisfaction Purchase Involvement Internet Experience Switching Barriers* Repurchase Intentions SEM / PLS Both switching barriers and overall N= 463 customers satisfaction had an equivalent from online retail positive direct effect on repurchase store in Taiwan. intentions. Community building had a dominant effect over other factors. Switching barriers are a stronger SEM predictor of repurchase intentions N= 526 online than satisfaction and community store consumers building is the stronger antecedent. Yang and Peterson, 2004 Perceived Value Customer Satisfaction: Switching Costs* Loyalty Chen and Hitt, 2002 Web site quality Product breadth Cost Personalization Web site ease of use Minimum deposit Switching Behaviour Logistic Regression N= 2,257 online consumers brokerage Industry Price increase Service quality Switching cost Customer lock-in Satisfaction Switching Barriers* Customer value Switching costs Satisfaction* Switching Intention SEM N= 520 Customers US Mobile Phone Company Moderating effects of switching barriers between customer satisfaction and switching intentions. Loyalty Interaction effect between customer satisfaction and switching costs on customer loyalty. Satisfaction Trust Switching Barriers Customer retention SEM N= 268 B2B customers of a courier service Hierarchic Regression N= 432 customers US telephone firm SEM N=287 credit car consumers and 288 long-distance consumers. MRA N= 228 Bank clients N= 206 hairstylist clients Online Context: Tsai and Huang, 2007 Tsai et al., 2006 Balabanis et al., 2006 Offline context: Shin and Kim, 2008 Lam et al., 2004 Ranaweera and Prabhu, 2004 e-Loyalty Burnham et al., 2003 Proced. Switching Costs Financial Switching Costs Relation Switching Costs Satisfaction Intention Stay Jones et al., 2000 Repurchase Intentions 166 Core-service satisfaction Switching barriers: Interpersonal relations Switching costs Attractiveness of alternatives Hierarchic Regression N= 192 Internet shoppers of most frequent site. SEM N= 235 consumers of online banking Conclusions Main effect and asymmetric interaction effect – the impact of switching barriers on e-store loyalty is greater when e-store satisfaction is low. Main effect switching costs were insignificant. Asymmetric effect of switching costs on loyalty is significant when customer satisfaction is above level. Switching behaviour is negatively correlated with high volume of web site usage, quality and breadth of offerings. Moderating effect of switching barriers between satisfaction and customer retention. Main effect - switching costs have a greater significant positive impact on customer retention than does satisfaction. Main effects of switching costs were insignificant. Asymmetrical interaction effect - switching barriers only have a significant positive impact on repurchase intentions when satisfaction is low. IADIS International Conference e-Commerce 2009 variance in consumers' intentions to stay with a current provider. This finding gives support to the theoretical assumption that switching costs are a stronger predictor of customer retention and this main effect is probably generated by market heterogeneity and product or service complexity. Provider heterogeneity is defined as the extent to which the providers in a market are seen as different or non-substitutable, increasing the uncertainty and the learning cost associated with switching providers (Burnham et al. 2003). Finally, when products or services are perceived as more complex (e.g. bank, credit services) or having higher purchase risk involvement (e.g. hairstylist) consumers are likely to perceive higher risks leading to uncertainty in product performance thus increasing the perceived economic cost. Finally, intangibility perceptions in the service industry imply higher risks and switching costs. Thus, consumers may rely more on relationships both with brands and with people to ensure that they receive a quality product and to simplify choices. 2.1.2 Main Effect of Switching Costs Consistent with Burnham et al. (2003) the main effect of switching costs explaining customer retention was again found by several researchers. Balabanis et al. (2006) surveying Internet shoppers of most frequent website, demonstrate that switching barriers explain more the variation in repurchase behavior than does satisfaction. Tsai et al. (2006) also found that perceived switching barriers from online stores are a stronger predictor of customer’s repurchase intention (.59) and the effect of satisfaction (.36) was again found to be much weaker than that of switching barriers. In a later study Tsai and Huang (2007) found that switching barriers were positively related to repurchase intentions from a particular e-retailer, and even had a direct effect that was equivalent to overall satisfaction. This result partially disagrees from previous studies. Actually, they found that overall satisfaction is positively related with switching barriers. This interrelationship between the constructs is supported by prior research and customer behaviour theory, which posit overall customer satisfaction with retailer's core-service or product performance increases customer dependence on a particular retailer and thus serves as an exit barrier. However, if a customer is not satisfied with the relationship with a provider, switching costs are generated. Furthermore, when a customer is satisfied with the relationship, the customer lacks motivation to seek alternatives. The lack of a main effect of switching costs observed may be due to other explanatory variables. According to Tsai and Huang (2007) community building has a dominant effect on repurchase intentions over other factors. Interestingly, the main effects of the switching costs were not significant on customer retention in several previous studies (Yang and Peterson, 2004; Jones et al., 2000). Also, these findings give support to the argument that switching costs are perceived to be lower in online environments. It is claimed that the lack of significant main effects for switching barriers do not, however, reduce their theoretical and practical importance because the interactions involving the barriers were significant (Baron and Kenny, 1986). In fact, Jones et al. (2000) found significant interactions indicating that there were effects of the switching barriers, but these effects only emerged as consumers became less satisfied with the coreservice offering. They argue that the absence of main effects only serves to reinforce their core thesis that a main effects approach is not sufficient to capture the complex processes underlying customer retention. 2.1.3 Moderating and Interaction Effects of Switching Costs Shin and Kim (2008) forecasting customer’s switching intention in mobile phone service, show that both satisfaction and switching barriers have a direct effect on switching intentions and also found the moderating effect of perceived switching barriers on the relationships between customer satisfaction and switching intentions. These results are consistent with previous studies, in a similar service industry. Ranaweera and Prabhu (2004) examining the influence of satisfaction, trust and switching barriers on customer retention, from a UK telephone company found that switching barriers have both a significant positive effect on customer retention as well as a moderating effect on the relationship between satisfaction and retention. Lam et al. (2004) in a B2B service context (corporate customers of courier service) examine the relationships and different strengths of satisfaction, value, switching costs and customer loyalty and find a positive interaction effect between customer satisfaction and switching costs on customer loyalty. An interaction effect (also know as an interactive effect) is similar to the moderator effect, and is the effect in which a third independent variable (the moderator variable) causes the relationship between a 167 ISBN: 978-972-8924-89-8 © 2009 IADIS dependent/independent variable pair to change, depending on the value of the moderator variable (Hair et al. 2006:172). In an online setting, Yang and Peterson (2004) on the basis of the aggregate sample found switching costs do not impose a significant direct or moderating effect on the association of customer loyalty with customer satisfaction and perceived value. Occasionally, a lower explanatory power of a variable may occur because of the large variation across individual observations when examining the aggregate sample (Pindyck and Rubinfeld, 1991). In fact, when examining sub-group analysis, Yang and Peterson’s study indicates an asymmetric interaction effect of switching costs. They observe that the moderating effect of switching costs only exists when customer satisfaction or perceived value level is above average. This finding is partially conflicting with theory, which posits that consumers only start to face switching costs when satisfaction with the provider starts to decrease. Regarding the interaction effect of switching costs, Balabanis et al. (2006) found the influence of perceived switching barriers on e-store loyalty is greater when e-store satisfaction is low. Contrarily to Yang and Peterson (2004) this finding is consistent with the interaction effects and the lack of main effects in Jones’s (2000) study, who concludes that only when satisfaction falls below a certain level do consumers begin to consider or be affected by the existence of switching barriers. This asymmetrical interaction effect under satisfaction level variation might be influenced by the extent to which switching costs are perceived in different service settings. Giving consistency to this finding, Jones et al (2002) conclude that the industry differences emerged across the switching costs dimensions. Specifically, the mean level of “pre-switching” search costs, evaluation costs, setup costs and perceived uncertainty were higher for hairstylists than banks, presumably due to their greater heterogeneity and intangibility. There is empirical evidence that the strength of switching costs on customer retention varies across industry (Burnham et al. 2003; Chen and Hitt, 2003) and is related to provider heterogeneity and intangibility. Moreover, the asymmetrical interaction effects found in Balabani’s study suggest a potential sample bias because more frequent customers might have higher perceived switching costs and this strength is enhanced online. Also, this may uncover a measurement error of satisfaction (e.g. use only one item) affecting the predictive power of the variable and model identification (Hair et al. 2006). 3. CONCLUSION We found conflicting results across empirical studies concerning the main effects as well as asymmetrical, moderating and interaction effects of switching costs between satisfaction and the loyalty relationship. The prevailing lack of main effects of switching costs on customer retention suggests that both satisfaction and switching costs have equal importance on determining customer retention (Tsai and Huang, 2007) and also reflects the strength of the interaction effect when satisfaction is above level (Yang and Peterson, 2004) or below level (Balabanis et al. 2006; Jones et al. 2000). These conflicting findings across studies deserve further investigation. The asymmetrical interaction effect subordinate to different satisfaction level variation on customer retention suggests that switching costs should be investigated under different satisfaction levels. While the studies analyzed provide strong support for the interaction effects of satisfaction and switching barriers on customer retention, they probably produce some biased results: Jones et al. (2000) analyze switching barriers and retention in a long-term contractual relationship with telephone service providers (the telephone industry has historically been closed to new entrants, making the industry a near monopoly) and besides, more frequent customers might have higher perceived switching costs (Balabanis et al., 2006). Apparently, the conflicting results concerning the effects of switching costs seem to be contingent to market and product heterogeneity or individual idiosyncrasies. In the service context there are higher switching costs (e.g. bank, telephone, credit car, insurance) due to service providers’ heterogeneity and customization. Moreover, the power asymmetries of switching costs, under satisfaction level variation on customer retention, apparently may uncover industry differences within an online setting (e.g.e-tail and online bank) and between an online versus offline bank service (Balabanis et al., 2006; Yang and Peterson, 2004; Jones et al., 2000). Further empirical validation of these effects in multiple settings would help shed further light on these and related phenomena of vital importance to different firms and marketplaces. It might be relevant to 168 IADIS International Conference e-Commerce 2009 investigate the effects of switching costs within a more competitive environment, such as the online retail sector, where the product’s standard, customer lower purchase involvement and almost zero search costs, decrease switching barriers. Future research should take this in consideration in order to avoid potential biased effects. Also, there is evidence that switching costs vary across industries: some researches found large firm-level variation in switching behaviour (Chen and Hitt, 2003). However, in an online context they are not deeply explored, especially in an online environment where switching costs are claimed to be lower (Balabanis et al., 2006). Furthermore, the mixed findings concerning the conflicting role of switching costs across studies may be due to a different measurement criterion. We claim that a more balanced set of measures is needed, allowing to reflect the multidimensionality of the construct which potentially increases its explanatory power. As switching costs have instrumental value to predict customer loyalty it is of paramount relevance to analyze how switching costs are measured and assess the effect sizes across studies. Conducting a Meta-Analysis of the reported findings is particularly desirable to identify the best measures for the switching costs construct and must be encouraged in future research. Finally, the present research can be extended in several ways. Other potential interaction effects between switching costs and relevant antecedents of satisfaction should be investigated, such as customer value and trust or community building, which potentially increases customer retention, specially towards an online retailer (Yang and Peterson, 2004; Tsai et al., 2006; Tsai and Huang, 2007). From a managerial standpoint, the interaction effects between switching costs and satisfaction on customer retention indicates that building both customer satisfaction and switching costs is a superior strategy rather to a focus on satisfaction alone to enhance customer retention and loyalty. Moreover, it is important to examine the implications of asymmetric interactions on investment decisions about switching costs or satisfaction. A strategy for firms to prevent customers switching is to create positive switching barriers that provide intrinsic benefits and create value for the customer (e.g. customer service, product offers, technology incentives). Actually, the strategic increase of artificial switching cost may boost unsatisfied customers and mask true loyalty. As customer retention is critical for a firm’s profitability in a competitive online market, managers should give top priority to develop strategies that both increase customer satisfaction and simultaneously switching costs to maintain customers in a long-term relationship. There is empirical evidence that switching barriers increase customer retention when satisfaction is low (Jones et al., 2000; Balabanis et al., 2006). However managers should be careful when creating switching barriers in lieu of satisfaction; this seems destined to failure in the long run particularly when (1) dissatisfaction is ongoing rather than temporary, and (2) the nature of the switching barriers are such that customers feel entrapped. Furthermore, as most of e-retailers are both “brick and click-and-mortar”, a clearer comparison between offline and online switching costs effects will help managers to strategically decide how to retain customers considering different retail formats. ACKNOWLEDGEMENT The authors thank the three Marketing Science Institute anonymous referees for helpful comments and suggestions provided which greatly improved the focus and clarity of this research. Ana Torres gratefully acknowledges the Ph.D grant from the Portuguese Foundation of Science and Technology. REFERENCES Book Besanko, David and Ronald Braeutigam, 2005. Microeconomics. John Wiley and Sons, Inc. Second Edition. East, Robert et al., 2008. Consumer Behaviour, Applications in Marketing. Kingston University, HEC Scholl of Management, University of South Australia. Hair, Joseph F. et al., 2006. Multivariate Data Analysis, Prentice Hall, 6/e, New Jersey, USA. Pindyck, Robert S. and Daniel L. Rubinfeld, 1991. Econometric Models and Economic Forecasts. McGraw-Hill International Editions, Economic Series, 4th Edition, Singapore. 169 ISBN: 978-972-8924-89-8 © 2009 IADIS Journal Anderson, Rolph E. and Srinivasan, Srini S., 2003. E-satisfaction and E-loyalty: A Contingency Framework. Psychology and Marketing, Vol. 20, Nº 2, pp. 99–121. Balabanis, George et al, 2006. Bases of e-Store Loyalty: Perceived Switching Barriers and Satisfaction. Journal of Business Research, Vol. 59, (February), pp. 214-224 Baron R. and Kenny, D. A., 1986. The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations. Journal of Personality and Social Psychology, Vol. 51, Nº 6, pp. 1173-1182. Burnham, A. et al, 2003. Consumer Switching Costs: A Typology, Antecedents and Consequences. Journal of the Academy of Marketing Science, Vol. 31, Nº 2, pp. 109–126. Chen, P., and Hitt. L., 2002. Measuring Switching Costs and the Determinants of Customer Retention in Internet-Enabled Businesses: A Study of the Online Brokerage Industry. Information Systems Research, Vol. 13, Nº 3, pp. 255–275. Chiou, J. S. 2004. The Antecedents of Consumers’ Loyalty toward Internet Service Providers. Information & Management, Vol. 41, pp. 685–695. Devaraj S. et al, 2002. Antecedents of B2C Channel Satisfaction and Preference: Validating e-Commerce Metrics. Information Systems Research, Vol. 13, Nº 3, pp. 316-333. Dick, Alan S. and Basu, K., 1994. Customer Loyalty: Toward an Integrated Conceptual Framework. Journal of the Academy of Marketing Science, Vol. 22, (Winter), pp. 99-113. Fornell, C. et al., 1996. The American Customer Satisfaction Index: Nature, Purpose, and Findings. Journal of Marketing, Vol. 60, pp. 7–18 Jones, M. A. et al, 2000. Switching Barriers and Repurchase Intentions in Services. Journal of Retailing, Vol.76, Nº 2, pp. 259–274. Jones, M. A. et al, 2002. Why Customers Stay: Measuring the Underlying Dimensions of Services Switching Costs and Managing their Differential Strategic Outcomes. Journal of Business Research, Vol. 55, pp. 441-450. Klemperer, P., 1987. Markets with Consumer Switching Costs. Quarterly Journal of Economics, Vol.102, Nº 2, pp. 375394. Lam, S. Y. et al., 2004. Customer Value, Satisfaction, Loyalty, and Switching Costs: An Illustration from a Business-toBusiness Service Context. Journal of the Academy of Marketing Science, Vol. 32, Nº 3, pp. 293–311. Mittal, Vikas and Kamakura, W. A., 2001. Satisfaction, Repurchase Intent, and Repurchase Behavior: Investigating the Moderating Effect of Customer Characteristics. Journal of Marketing Research, Vol. 39, (February), pp. 131–42. Oliver, Richard L., 1999. Whence Consumer Loyalty? Journal of Marketing, Vol. 63, pp. 33–44. Ranaweera, C. and Prabhu, J., 2004. The Influence of Satisfaction, Trust and Switching Barriers on Customer Retention in a Continuous Purchasing Setting. International Journal of Service Industry Management, Vol. 14, Nº 4, pp. 374395. Reichheld, Frederick F. and Schefter, P., 2000. E-loyalty: Your Secret Weapon on the Web. Harvard Business Review, July–August, pp. 105–113. Shankar, Venkatesh et al., 2003. Customer Satisfaction and Loyalty in Online and Offline Environments. International Journal of Research in Marketing, Vol. 20, pp. 153–175. Shin, D-H and Kim, W-Y., 2008. Forecasting Customer Switching Intention in Mobile Service: An Exploratory Study of Predictive Factors in Mobile Number Portability. Technological Forecasting & Social Change, Vol.75, Nº 6, pp. 854874. Srinivasan, S., R. et al., 2002. Customer Loyalty in e-Commerce: An Exploration of its Antecedents and Consequences. Journal of Retailing, Vol.78, pp. 41–50. Szymanski, D. M. and Henard, D. H, 2001. Customer Satisfaction: A Meta-Analysis of the Empirical Evidence. Journal of the Academy of Marketing Science, Vol. 29, Nº 1, pp. 16-35. Tsai, Hsien-Tung and Huang, Heng-Chiang, 2007. Determinants of Repurchase Intentions: An Integrative Model of Quadruple Retention Drivers. Information and Management, Vol. 44, pp. 231-239. Tsai, Hsien-Tung, et al., 2006. Why On-line Customers Remain with a Particular E-Retailer: An Integrative Model and Empirical Evidence. Psychology and Marketing, Vol. 23, Nº 5, pp. 447-464. Yang, Z. and Peterson, R. T., 2004. Customer Perceived Value, Satisfaction, and Loyalty: The Role of Switching Costs. Psychology and Marketing, Vol. 21, Nº 10, pp. 799-822. Zeithaml, V. A. et al., 2002. Service Quality Delivery Through Web Sites: A Critical Review of Extant Knowledge. Journal of the Academy of Marketing Science. Vol. 30, Nº 4, pp. 362–375. 170 IADIS International Conference e-Commerce 2009 RECOGNITION OF EMOTIONS IN E-COMMERCE-SETTINGS Prof. Dr. Susanne Robra-Bissantz Technische Universität Carolo-Wilhelmina zu Braunschweig, Institut für Wirtschaftsinformatik, Abteilung Informationsmanagement ABSTRACT If the recognition of emotions in e-Commerce-Settings was possible, webshops could respond to customers’ emotions just like a personal seller. This would potentially increase the share of fully completed shopping processes. The following contribution introduces concepts, experiments and first results of recognition of customers’ emotions in e-Commerce. KEYWORDS E-Commerce, Emotion Recognition, Experiment, Individualisation, Biometrical Sensors 1. MOTIVATION, IDEA AND OBJECTIVE In traditional shopping processes the personal seller responds intuitively and individually to his customer. This also refers to the customer’s emotions, such as anger or joy, when he is offered a product. Although eCommerce should be seen as an electronic interactive selling process, the consideration of customers’ emotions is missing. This leads to the situation that an angry customer quits his shopping basket or a helpless customer leaves the shop after an unsuccessful search for a product. If it was possible to automatically record customers’ emotions in e-Commerce and to react to these automatically as well, the number of fully completed shopping processes would most probably increase (Robra-Bissantz and Zabel, 2005: 168–175). The basic idea of this contribution is, that emotions can be recognized by the help of biometrical data of human beings like pulse or blood volume pressure, that are intelligently processed. The measuring of biometrical data needs sensors that may be integrated in a computer mouse or a mobile phone. Still for the use case of e-Commerce no intelligent and reusable tool for processing emotions from biometrical data exists. Therefore the first steps towards recognition of emotions in e-Commerce are experiments that enable the implementation of such a tool. The following article introduces a set of experiments that lead to an IC-System which may recognize emotions from biometrical data. It starts with a short overview of the findings of Affective Computing, which deals with the consideration of emotions in IC-Systems. Affective Computing is the basis of recognition of emotions in e-Commerce. Afterwards the psychological construct of emotion will be introduced. This leads to different possibilities of how to measure it. The usage of biometrical data in order to measure emotions is the one possibility that can be used in e-Commerce, respectively over a distance. Other possibilities to measure emotions are applicable in our experiments in order to validate emotions. Then Neural Networks and Fuzzy Systems are introduced as means of an intelligent processing of biometrical data to emotions. Different possibilities to validate emotions in experiments and different ways to process biometrical data lead to different experimental settings. The following execution of experiments shows their advantages and disadvantages as well as first results concerning the performance of machine aided emotion recognition. 2. EFFECTIVE COMPUTING Human beings react to computers very emotionally and also expect them to be emotional as well (Reeves and Nass, 1996: 253). Today’s computers however are not able to deal with emotions. They only use single 171 ISBN: 978-972-8924-89-8 © 2009 IADIS aspects of the possibilities of human thinking. The so called trilogy of mind comprises of cognition, motivation and emotion (Buck, 1988: 3). Only parts of cognition are implemented in computers. Motivation and emotion are hardly considered, although it is obvious, that human thinking strongly depends on emotions as well (Picard, 2000: 11; Goleman, 1998: 75). Approaches that try to solve this problem are summed up by the term „Affective Computing“ (Picard, 2000). It deals with the recognition of emotions as well as with the notion of enabling computers for emotional reactions. The overall objective of Affective Computing is the integration of emotions into the information processing of computers, which is supposed to lead to better computerized decisions (Picard et al., 2001: 1175- 1191). For this purpose the computer reproduces internal, affective processes of human beings. Like human attention it senses emotional and cognitive signals (signal input). Via decoding and processing an impression and reactions emerge. In this process cognitive and affective dispositions (like knowledge, experiences or attitudes) play an important role. Reactions are coded into a signal output that corresponds to human behavior. With computers these processes do not work automatically, but they have to be implemented step by step: • Signal Input (Emotion Measurement): the problem in measuring emotions is that input channels of a computer are normally restricted to keyboard and mouse. Human beings are able to sense more information than just the content of a message, e. g. from the inflection, the gestures or the facial expression of the sender. Affective Computing has therefore developed different additional input channels, such as Affective Tangibles, that may perceive physical reactions. Furthermore biometrical data are measured, speech, inflection or facial expressions are analyzed (Ark et al., 1999: 818–823). • Decoding (Emotion Recognition): In order to decode input data from different channels, Affective Computing mainly uses stochastic methods that lead to the most likely emotion (Burleson and Picard, 2004). • Processing and Signal Output: A computer uses models about human behavior in order to work out an emotional reaction. These models are kept in a database, just like the human inherently memorizes his dispositions. As today’s main areas of Affective Computing are learning or social issues, like drug counseling, the main theoretical models used up to now are out of hese areas, like learning models (D'Mello et al., 2005: 7-13). For emotional reactions Affective Computing uses avatars or robots that use different facial expressions or gestures in order to express emotions. In part Emotion Recognition in e-Commerce may use the findings of Affective Computing. But there are a couple of differences: First of all, it is no objective in e-Commerce that computers are really able to be emotional. Therefore research is primarily restricted to the recognition of emotions. Secondly, in general the requirements in e-Commerce are lower in terms of correctness of the interpretation of sensed emotions. At first it would be enough to roughly detect striking emotions. Therefore the innovative software solution of soft computing will be used instead of stochastic methods. Thirdly, in e-Commerce it must be possible to measure emotional signals over a distance and electronic networks. Therefore biometrical data and corresponding sensors that may be integrated in a mouse or a mobile phone are applied for input purposes. 3. EMOTIONS 3.1 Psychological Construct Emotion is a manifold phenomenon (Desmet, 2003: 112), with no final and successful explanation model yet. In times of behaviorism it was even doubted that emotion existed as an independent concept (Holzer, 1987). The complexity of emotion results from the „indescribably complex” interlacing of simple basic emotions (Minsky, 1986: 328). From today’s standpoint of psychology emotion possesses four dimensions (Kroeber-Riel and Weinberg, 2003: 105). These are arousal, valence, quality and consciousness. The first two dimensions describe intensity and assessment (positive or negative) of a feeling. The third dimension, quality, expresses personal perception and evaluation of an emotion. It is highly subjective and differs strongly between different human beings. Quality makes it very hard to measure emotions in general. The fourth dimension, consciousness, leads to emotions that are only figuratively but not verbally conscious (Kroeber-Riel and Weinberg, 2003: 106). This dimension complicates questioning as a measurement method of emotions (Plutchik, 1991: 22). 172 IADIS International Conference e-Commerce 2009 Human beings comprise of at least six innate primary emotions, independently from their culture or language: joy, surprise, fear, sadness, anger and disgust (Plutchik, 1991:61; Izard, 1992: 561–565; Camras et al., 1997: 301; Ortony and Turner, 1990: 315–323). During adolescence cognitive emotion processing arises, that increasingly covers primary emotions. Primary emotions may be arranged like a chromatic circle (Plutchik, 1991: 109). Similar emotions are placed close to each other and their mixture leads to so called secondary emotions. Like this the whole spectrum of emotions can be depicted (Plutchik, 1991: 115 et seq.). 3.2 Emotion Recognition – Measuring Methods There are different approaches of emotion measurement with different advantages and disadvantages. Basically they can be classified into methods of self report and methods of direct observation of spontaneous bodily reactions. Self report always runs the risk of cognitive tampering, e. g. for reasons of social desirability (Kroeber-Riel and Weinberg, 2003: 104). Observations feature the disadvantage that the connections between bodily reaction and emotion have to be learnt at first (Darwin, 1964; Grings and Dawson, 1978: 3). 3.2.1 Self Report Self report works for example with a questionnaire that raises emotions after their cognition, with verbal annotations during emotion perception as well as with slide controls or turning knobs, which can be used to sort of tune in emotion while it is sensed. Self report during perception has the advantage that emotions are expressed more spontaneously during their perception than afterwards. But a Self report with an additional gadget features the disadvantage that it disturbs the proband’s concentration. More often than not gadgets are also much too complicated to be handled while an emotion arises. A useful and only little disturbing gadget is a turning knob that only characterizes emotions by their valence and arousal (Cowie et al., 2000). These dimensions are tuned in by turning the knob more or less (arousal) to the right or left (valence: positive or negative). 3.2.2 Spontaneous Bodily Reaction Bodily reactions that enable the recognition of emotions are changes of facial expressions or of biometrical features such as pulse or skin resistance. Table 1. FACS Coding emotion FACS action units surprise fear disgust eyebrows lifted, eyes wide open, jaw dropped eyebrows lifted and knitted, eyes wide open, but lower lid drawn upwards, mouth open nose wrinkled up, eyebrows drawn downwards, upper lip drawn upwards, lower lip against upper lip or downwards and forward, cheeks drawn upwards eyebrows drawn downwards and knitted, eyes knitted and fixed, mouth closed and lips pressed together corners of the mouth drawn up- and backwards, mouth closed or open or even visible teeth, cheeks drawn upwards eyebrows knitted upwards, lower lid drawn upwards, glaze downwards, corners of the mouth drawn downwards anger joy sadness An analysis of emotions from facial expressions is complicated, as facial muscles are not only used for the expression of emotions but also for conscious reactions (Bänninger-Huber, 1985: 36 et seq.). FACS (Facial Affect Scoring Technique) is a technique that supports this analysis and is itself supported by fitting software (Ekman and Friesen, 1984). FACS divides up facial expressions in smallest atomic activities of facial muscles (action units). These action units can be combined to patterns for different emotions (Bänninger-Huber, 1985: 49; Kroeber-Riel and Weinberg, 2003: 113). With FACS a mimic is firstly described by a combination of action units. Only afterwards it is interpreted as an emotion with FACSAID 173 ISBN: 978-972-8924-89-8 © 2009 IADIS (Ekman and Rosenberg, 1997), a database that assigns emotions to patterns of action units. Table 1 shows, how primary emotions show up as a mimic (Ekman, 1999: 13). Table 2. Recognition of Emotion from Biometrical Data biometrical feature / measurement pulse: EKG blood pressure: mmHg blood volume pressure skin conductance: mircromho muscle potential: EMG (head) temperature: °C at the finger breathing: frequency and depth emotion little increase: joy, disgust, surprise, strong increase: anger, fear, decrease: sadness increase: fear, anger, sadness change: surprise, fear decrease: disgust, increase: fear, surprise, arousal (joy: no change) short and strong increase: surprise, increase: joy, anger, fear, arousal increase: anger, surprise, quick changes: arousal, (joy: no change) little increase: joy, anger, surprise, sadness, fear For the recognition of emotions from biometrical data a mature technique, such as FACS, does not exist. But at least devices are available, that use biometrical data for a so called Biofeedback, the feedback of physical functions (e. g. SOFTmed Physio-System, Insight Instruments (Insight Instruments, n.d.)). Suitable for the measurement of emotions are pulse, blood pressure, blood volume pressure, skin resistance (skin conductance), body temperature, breathing and muscle potential. The latter two features are controlled by the central nervous system and are therefore more often consciously formed than the others (Grings and Dawson, 1978: 12). Table 2 shows biometrical features, their measurement as well as first indicators for emotions, drawn from literature (Grings and Dawson, 1978: 12). Measuring biometrical data is an approach that seems to be suitable for e-Commerce. All necessary data can be identified, automatically, noninvasively and only almost nonrestrictively. Sensors for most biometrical data may be integrated into a mouse (Grings and Dawson, 1978: 12; Holzer, 1987; Baltissen, 1983; Muthny, 1984: 49; Carlson and Hatfield, 1991: 240; Schandry, 1989; McNaughton, 1989). The analysis of facial expressions as well as different forms of self reports is suitable for an experimental setting, where an ICsystem shall be designed, that recognizes emotions from biometrical data. 3.3 Emotion Recognition – Processing Methods Intuitively, a Neural Network appears suitable if emotions are to be calculated, as Neural Networks are designed to imitate human information processing. Additionally another Softcomputing approach is tested: Fuzzy Logic. Both approaches will use biometrical data as input vector in an IC-System for emotion recognition. From biometrical data the IC-System will calculate the emotion of the human being. Both Softcomputing methods cannot be programmed. They learn the relationship between input and output through learning with existing data vectors. Therefore experiments are needed, that produce an input vector consisting of biometrical data as well as the corresponding output vector consisting of emotions. These emotions will be measured by the means of self report or analysis of facial expressions. Neural Networks and Fuzzy Systems will be trained until they are able to calculate emotions from input vectors with unknown output vectors. 3.3.1 Neural Networks A Neural Network (NN) is a computational model based on biological neural networks. It consists of an interconnected group of artificial neurons (simple processing units) and processes information from an input layer through hidden layers to an output layer. Configuration and functions of a neural network have to be adapted to the task by the trainer. For emotion recognition the engineering platform „Stuttgart Neural Network Simulator” is used (Zell, 1996). This tool requires a couple of decisions. • Network topology: The topology of the neural network is responsible for the propagation of signals between neuron layers and within. The well known Feedforward networks only forward signals upstream from layer to layer. This topology does not stipulate timely dependencies of the input vectors. As emotions do not emerge independently from former emotions, the so called Elman topology is used. Elman networks 174 IADIS International Conference e-Commerce 2009 implement additional context neurons into Feedforward networks that introduce the output of the previous cycle as additional input for the next cycle (Elman, 1990: 179-211). • Input vector: It has to be avoided, that the neural network can approximate training data perfectly but is not able to work with new input data (over fit). Therefore data from experiments is divided up into training data and validation data. Additionally Neural Networks with many input data are often less generizable. Therefore only those biometrical data are used, that show a significant explanatory contribution (Rojas, 1996: 144). • Output vector: In e-Commerce settings it is possible to reduce the output vector to lesser emotions (e. g. valence / arousal), if the neural network is not able to approximate all primary emotions exactly enough. • Number of hidden layers and their neurons: Neurons in hidden layers use an activation function in order to transform the input vector into the output vector. The number of hidden layers and their neurons has to be simulated (Sarle, 2002: 3). As activation function the neurons use Tangens Hyperbolicus, which converges faster than a Sigmoid function (Bishop, 1995: 127). • Learning paradigm: Neural networks learn by adapting the weights of the connections between neurons of different layers. During learning the well known backpropagation algorithm shows a couple of problems (Zell, 1996). Only a very little learning rate may avoid, that this algorithm is stuck in local optima of the error area. But a higher learning rate may oversee global optima. Therefore a so called momentum is introduced. It increases the learning rate if the algorithm works on a plateau of errors and decreases it, if the error changes rapidly. The resulting learning method uses a Backpropagation-Momentum approach with simulations along different learning rates and momentums. 3.3.2 Fuzzy-System Fuzzy logic approaches may be used for basically all mathematical functions. Here a Fuzzy Expert System for Emotion Recognition is implemented with XFuzzy, an Open Source Fuzzy Shell (Fuzzy Logic Design Tool XFuzzy 3.0, n.d.). In a Fuzzy Expert System a membership function, with fuzzy sets that range between 0 and 1 transforms sharp input and output variables into rather approximate linguistic variables. A rule base consists of rules that transform these input variables into the output variables. Membership functions with different parameters as well as the rule base have to be built by the trainer. First membership functions use well known forms. First rules are derived from former findings about the correlation between biometric data and emotions (see table 1). All parameters of the membership functions and rules are improved with the input and output vectors from the experiments in different simulations. • In- and output variables: In- and output variables have to be reduced significantly, compared to the Neural Network, as the whole rule base has to be built manually. In e-Commerce settings it might e. g. be reasonable, to choose only the one emotion anger. For input variables those four are chosen, that correlate strongest with the output variable. They are at first coded as linguistic variables with five values and with triangular membership functions (cf. Table 3). Table 3. Fuzzy Sets – Limits (u. lim. : upper limit, l. lim. : lower limit) fuzzy set variable EMG 1 pulse amplitude skin conductancy temperature anger very high high medium low very low u. lim. l. lim. u. lim. l. lim. u. lim. l. lim. u. lim. l. lim. u. lim. l. lim. 155.26 6.18 6.18 4.58 4.58 3.74 3.74 3.14 3.14 2.11 158.50 70.17 70.17 35.18 35.18 19.68 19.68 10.32 10.32 0.00 5.52 2.53 2.53 2.31 2.31 2.06 2.06 1.11 1.11 0.81 32.59 31.95 31.95 31.42 31.42 30.25 30.25 27.50 27.50 25.01 2.0 1.6 1.6 1.2 1.2 0.8 0.8 0.4 0.4 0.0 0 0 0 0 0 0 0 0 0 0 • Improvement in XFuzzy: In first simulations different gaps in the rule base are identified and filled with additional rules. A further improvement works with simulations of form and position of membership fuzzy sets and with operators within as well as between the rules. Learning at work: With an own developed Tool (FuzzyEmotion) the trainer may change single inputoutput-correlations via drag&drop-operations also after the training. He may, for example, assign another 175 ISBN: 978-972-8924-89-8 © 2009 IADIS value for the variable anger to specific input values, according to new experiments. This new output value initiates another automatic improvement of the whole Fuzzy System. 4. EXPERIMENTS A couple of experiments with similar settings reveal biometric data as well as statements about really sensed emotions that are as objective as possible. They serve as input and output vectors for the training of the Neural Network and the Fuzzy System. For these experiments a measurement device for biometrical features with different sensors (Biofeedback-System, see chapter 2) and a setting for self reports or the recognition of facial expressions is needed. Additionally, the experiment works with different stimuli that provoke the emotions of the probands (see Figure 1.). Emotion Measurement Biometrics Stimulus Emotion Emotion Validation Facial Recognition Self Report Emotion Recognition Training with Input (Emotion Measurement) and Output (Emotion Validation) Figure 1. Experiments for Emotion Recognition Stimuli are meant to provoke emotions of probands. For experiments with primary emotions the best stimuli, according to psychological studies, are short samples of well known movies or videos. They lead to stronger emotions than pictures and also than commercials. In order to provoke the emotion anger, a manipulated computer game (Tetris) can be used (Holzer, 1987; Niemann, 2004). A break down of the game provokes the strongest emotions. The stimulus “webshop” takes into account that the planned application area of emotion recognition is meant to be e-Commerce. During experiments it either breaks down, like the Tetris play or the proband is faced with a complex up to impossible shopping task. In order to spot position and behaviour of the proband in the webshop, an IC-System was developed (E3Blogger). It registers all his keyboard and mouse movements. For self reports the experimental setting either uses a turning knob or a questioning of the proband after the experiment. The only problem is, that then he is often unable to remember his emotions in single situations, e. g. during the shopping process. Therefore a split screen is used, where the proband can see a video of his face during the experiment and at the same time a video of the stimuli, e. g. of his website usage. The video of the proband’s face can also be used for the recognition of his facial expressions. Table 4 shows the experimental settings of four different experiments for emotion recognition. Table 4. Experiments - Overview experiment E1: „ambitioned“ stimulus film, Tetris E2: „anger“ E3: „e-Commerce“ Tetris Webshop E4: „optimized“ film samples validation facial expressions, questionaire E3Blogger E3Blogger, questioning, split screen turning knob processing Neural Network output primary emotions Fuzzy System Fuzzy System anger anger / joy Neural Network valence, arousal 4.1 Specialities and Results The group of experiments shows first results but also the learning process of the conductor. Planned settings sometimes had to be changed because of completely unexpected problems. 4.1.1 Experiment E1: „ambitioned“ E1uses the original experimental setting, derived from literature on Affective Computing and from psychological experiments. The probands (32 Persons of different age and gender) were connected to biometrical sensors and shown different film samples that seemed suitable to provoke different primary 176 IADIS International Conference e-Commerce 2009 emotions (e.g. Lord of the Rings). These were validated during the experiment with a little questionnaire: „please tick the emotion, felt during the film“. Additionally the facial expression of the proband was filmed during the experiment and later analyzed with the FACS System. More than 14.000 data sets resulted, that assign biometrical input data to „objective” emotion. With these data sets a Neural Network was trained as this experiment comprises too many variables for a fuzzy system. A first result was that especially male probands hardly show any changes of facial expressions. The reason is probably that the camera, positioned well seen, worked as a social control that prevented obvious emotions. Therefore only the intended emotions of the film samples and the results of the questioning are available as output variables for the training of the Neural Network. Simulations lead to the best results with Neural Networks with two hidden layers, with the Elman algorithm and small learning rates. Here the MSE (Mean Square Error) between the output values from the experiment and the calculated output values is in the interval [0.15, 0.53]. But still a reference value, the MSE of a network with constant input could not be achieved. This is probably because the Elman network literally learns values from former training cycles by heart. Therefore different simulations with two or only one output (valence and arousal or anger) were run with feedforward networks. Here the results are quite encouraging, as the outputs can be learnt almost faultless. But still only in some configurations the MSE is below its reference value. The first experiments of emotion recognition therefore proof the method to be principally suitable. But the approach of measuring eight primary emotions seems to be too ambitious. Also the stimulus film samples and the validation only with a questioning of emotions does not lead to the hoped for success. The emotions that were felt by the probands could not be measured exactly enough. Additionally biometrical data were measured during the whole film sample whereas the emotions most probably only arise in short sequences. Valid data are therefore probably falsified by the load of inaccurate datasets. 4.1.2 Experiment E2: „anger“ With the experiences from E1, the second experiment only tries to recognize the emotion anger. Additionally the stimulus was changed from film samples to the computer game Tetris that was manipulated and crashed by the conductor. 28 persons of different age and gender were asked to play Tetris while being connected to the biometric sensors. Their keystrokes were recorded as well as their face and the Tetris screen. This setting is suitable for a training with a Fuzzy System. Simulations show that this Softcomputing method even leads to better results for this experiment. The rules of the Fuzzy System are shown in Table 5. They result from the assumptions from chapter 3.2 and from a regression of the input and output values from the experiment. In different trainings data are aggregated or processed individually, rules are derived from a linear or polynomial regression and “and-“ or “or-“ operators are used for the conjunction of rules. Table 5. Rules of Fuzzy-Systems skin conductance EMG temperature pulse amplitude anger very high very high very low, very high very high very high very high, medium, low high, medium, low medium, low, very low high, medium low high low medium, low high high very high, high, medium high, medium, very low high low very low very high medium high very low medium In discussions after the experiment single probands had the possibility to improve the fuzzy system by adapting the calculated output value to the really felt emotions with the System FuzzyEmotion. As result the simulations with learning data reach a MSE of 0.02 to 0.05. The MSE with validation data lies between 0.07 and 0.10. The total variance of the variable anger (interval [0, 2]) is 0.15. A learning MSE of 0.035 therefore means a not declared variance of 23.3%. With validation data and its MSE of 0.07 the mean deviation from the real output value is 0.26. Best results could be reached with a system that is based on a linear regression between in- and output and, quite significantly, when the rule base is adapted by probands. The results of this experiment seem better as the ones from E1. Reasons are potentially the reduction of the output to one single variable “anger”, a better stimulus and the use of a Fuzzy System with a manual aggregation of the results to a rule base. 177 ISBN: 978-972-8924-89-8 © 2009 IADIS 4.1.3 Experiment E3: „e-Commerce“ In a next experiment the emotion anger is provoked with a complex task in a webshop. In order to make the probands feel unobserved, cameras are hidden and the conductor leaves the room. After the experiment the probands are questioned about their emotion with the split-screen setting. For this questioning the conductor picks situations, where the biometric data and their processing with the Fuzzy System of E2 show the emotion anger (Kammerer, 2005). Experiments with 20 probands of different gender and age lead to 97 of these situations. 54 of these situations really resulted from anger or, more generally, negative emotions. Another 31 situations however resulted from positive emotions. Thus probands did really feel an emotion in most detected situations. Contrary the probands claimed, that they in sum only remembered another two situations, where they felt anger that was not noticed by the biometrical data. In sum the results of E3 show that the Fuzzy System does register the arousal of probands. The valence of an emotion (positive or negative) however cannot be interpreted exactly enough. 4.1.4 Experiment E4: „optimized“ E4 uses the experiences from all other experiments. It uses a smaller number of output variables, better stimuli and a new way of validation of emotions. Still it tries to come to recognition of quasi all emotions by using valence and arousal. Stimuli are much shorter sequences (20 to 30 seconds) of daily news, quiz- and game shows or short films. The proband expresses his emotions during the experiment with the turning knob. As Softcomputing system the Neural Network is used, according to the experiences from E3. Simulations are carried out with data from 22 probands of different age and gender (more than 12.000 data sets). Best results can be achieved with Feedforward networks with four layers and a relatively high number of neurons in the hidden layers. While increasing the number of these neurons stepwise up to 50 the learning mistake declines until a MSE of 0.05. As the total variance of the output is 0.5, this means a not declared variance of 10%. The validation error reaches its minimum with Neural Networks with 25 neurons in the hidden layers (MSE = 0.19, not declared variance 37%). 5. CONCLUSION The recognition of emotions in e-Commerce-Setting is an ambitious goal. Still first results from experiments lead to the conclusion that it is definitely possible. But there is a long way to go until hardware, like an emotion-mouse, is broadly available and until a recognition system exists, that can be easily used for every webshop and that leads to appropriate reactions on human emotions. In closer future the recognition of emotions can nevertheless be used for the test of websites. This leads to new findings during website evaluation and additionally, quasi as a side effect, to new testing and training data for an improvement of the emotion recognition system. Therefore a flexible platform was implemented, that makes it easier to carry out experiments and also website tests. It allows the management of experiments and single tests and offers import and export interfaces to different systems (e. g. the presentation of the stimulus, the turning knob, the biofeedback system, the Neural Network and the Fuzzy System). All data are integrated in records with time stamps (Hausmann, 2006). REFERENCES Ark, W. S. et al, 1999. The Emotion Mouse. In Proceedings of the 8th HCI International on Human-Computer Interaction. Lawrence Erlbaum Associates, Mahwah, pp 818–823. Baltissen, R., 1983. Psychische und somatische Reaktionen auf affektive visuelle Reize bei jungen und alten Personen. dissertation, University of Düsseldorf. Bänninger-Huber, E., 1985. FACS (Facial Action Coding System) als Instrument zur Erfassung therapeutischer Veränderungen. dissertation, Zürich. Bishop, C. M., 1995. Neural Networks for Pattern Recognition. Oxford University Press, Oxford. Buck, R., 1988. Human Motivation and Emotion. John Wiley & Sons Inc, New York. Burleson, W. and Picard, R. W., 2004. Affective Agents: Sustaining Motivation to Learn Through Failure and a State of Stuck [online]. Available at: http://affect.media.mit.edu/pdfs/04.burleson-picard.pdf [Accessed 9th August 2007]. 178 IADIS International Conference e-Commerce 2009 Camras, L. A. et al, 1997. Japanese and American Infants’ Responses to Arm Restraint, In Ekman, P. and Rosenberg, E. (ed.), What the Face Reveals. Oxford University Press, New York, pp 289–301. Carlson, J. G., Hatfield, E., 1991. Psychology of Emotion. Harcourt Brace College Publisher, Philadelphia. Cowie, R. et al, 2000. ‘FeelTrace’, an instrument for recording perceived emotion in real time [online]. Available at: http://www.dfki.de/~schroed/articles/cowieetal2000.pdf [Accessed 28th December 2007]. D'Mello, S. K. et al, 2005. Integrating Affect Sensors in an Intelligent Tutoring System. In Proceedings of the Affective Interactions: The Computer in the Affective Loop Workshop at 2005 International conference on Intelligent User Interfaces, ACM, New York, pp 7-13. Darwin, C., 1964. Der Ausdruck der Gefühle bei Mensch und Tier. Rau, Düsseldorf. Desmet, P., 2003. Measuring Emotions: Development and Application of an Instrument to Measure Emotional Responses to Products. In Blythe, M. A. et al. (ed.), Funology: From Usability to Enjoyment. Springer, Dordrecht, p 111-123. Ekman, P., 1999. Facial Expressions. In Dalgleish, T. and Power, M. (ed.): Handbook of Cognition and Emotion. Wiley, Sussex, p 13. Ekman, P. and Friesen, W. E., 1984. Unmasking the Face. Malor Books, Palo Alto. Ekman, P. and Rosenberg, E. (ed.), 1997. What the Face Reveals. Oxford University Press, New York. Elman, J. L. 1990. Finding Structure in Time. In Cognitive Science No. 14, pp 179–211. Available at: http://homepages.inf.ed.ac.uk/keller/teaching/connectionism/CogSci90-Elman.pdf [Accessed 21st September 2007]. Fuzzy Logic Design Tools XFuzzy 3.0 [online]. Available at: http://www.imse.cnm.es/Xfuzzy/Xfuzzy_3.0/Xfuzzy3.0_en.pdf [Accessed 1st November 2004]. Goleman, D., 1998. Emotionale Intelligenz. Dtv, München. Grings, W. W. and Dawson, M. E., 1978. Emotions And Bodily Responses – A Psychopsychological Approach. Academic Press, New York. Hausmann, S., 2006. E3-Studio: Eine Experimentierumgebung zur Erkennung von Emotionen im E-Business. student research project, University of Nürnberg. Holzer, H., 1987. Psychophysiologische Analyse emotionaler Situationen. dissertation, University of Ulm. Insight Instruments: SOFTmed Physio-System [online]. Available at: http://www.insight.co.at [Accessed 1st June 2007]. Izard, C. E., 1992. Basic Emotions, Relations Among Emotions and Emotion–Cognition Relations. In Psychology Review Vol. 99, No. 3, pp 561–565. Kammerer, S., 2005. Konzeption und Validierung von Regeln zur Erkennung von Emotionen aus dem Klickverhalten eines Internetnutzers. student research project, University of Nürnberg. Kroeber-Riel W. and Weinberg, P., 2003. Konsumentenverhalten. Vahlen, München. McNaughton, N., 1989. Biology and Emotion. Camebridge University Press, Cambridge. Minsky, M. L., 1986. The Society of Mind. Simon & Schuster, Touchstone. Muthny A. F., 1984. Elektrodermale Aktivität und palmare Schwitzaktivität als Biosignale der Haut in der psychophysiologischen Grundlagenforschung. dissertation, University of Freiburg, Freiburg. Niemann, C., 2004. Konzeption und Realisierung der Komponente zum Emotionsmonitoring eines Effizienzevaluationsplatzes für E-Business-Anwendungen (E3). dissertation, University of Nürnberg. Ortony, A. and Turner, T. J., 1990. What’s Basic About Basic Emotions?. In Psychological Review Vol. 97, No. 3, pp 315–323. Picard, R. W., 2000. Affective Computing. The MIT Press, Cambridge. Picard, R. W. et al, 2001. Toward Machine Emotional Intelligence: Analysis of Affective Physiological State. In IEEE Transactions Pattern Analysis and Machine Intelligence, Vol. 23, No. 10, pp 1175- 1191 Plutchik, R., 1991. The Emotions. University Press of America, Lanham. Reeves, B. and Nass, C., 1996. The Media Equation. Cambridge University Press, Stanford. Robra-Bissantz, S. and Zabel, A., 2005. Push Concepts to Control Interactions in E-Commerce. In CEC IEEE 2005: Seventh IEEE International Conference on E-Commerce-Technology. IEEE Computer Society Press, Los Alami, pp 168–175. Rojas, R., 1996. Neural Networks – A Systematic Introduction. Springer, Berlin. Sarle, W. S., 2002. Neural Network FAQ.. [online]. Available at: ftp://ftp.sas.com/pub/neural/FAQ.html [Accessed 21th September 2007]. Schandry, R., 1989. Lehrbuch der Psychophysiologie: Körperliche Indikatoren psychischen Geschehens. BeltzPVU, München. Zell, A., 1996. Simulation Neuronaler Netze. Oldenbourg, Bonn. 179 Short Papers IADIS International Conference e-Commerce 2009 E-PROCUREMENT ADOPTION BY SUPPLIERS: A RESEARCH PROPOSAL Paulo Andrade ISCTE – Lisbon University Institute Av. das Forças Armadas, 1600-082 LISBOA – Portugal Bráulio Alturas ADETTI/ISCTE ISCTE – Lisbon University Institute Av. das Forças Armadas, 1600-082 LISBOA – Portugal ABSTRACT This paper presents a current research with the aim to identify the enablers, barriers and possible solutions to eprocurement adoption by suppliers in Portugal. The literature relating to e-procurement implementation and operation is reviewed, with focus on the barriers and enablers already identified in the literature. A research methodology is proposed to study the problem, and this work will contribute to better address the issues faced by suppliers on e-procurement implementations. KEYWORDS E-Procurement, Collaboration, E-Commerce, B2B, Supply Chain Management. 1. INTRODUCTION E-procurement is the generic term applied to the use of integrated database systems and wide area (commonly web-based) network communication systems in part or all of the purchasing process (Croom & Brando-Jones, 2005). E-procurement allows buyers to automate transactions and focus on more strategic activities. E-procurement solutions also contribute to a better organizational performance, allowing reductions in cost and time when ordering from suppliers, and helping to achieve a well-integrated supply chain. Although there are many benefits in e-procurement solutions, there also appears to be some barriers to their successful implementation. Former research shows that many companies still prefer the traditional methods to communicate and exchange with business partners. Companies need to understand better how to implement e-procurement solutions on an efficient and effective manner. Any successful e-procurement system needs suppliers that are willing and able to trade electronically and their co-operation is crucial to the project's success. This degree of openness and transparency is new to most organizations, and it requires relevant cultural changes and high levels of trust between the participants (Harris & Dennis, 2004). This paper presents issues related to supply chain management, e-commerce, e-procurement and the barriers and enablers for supplier adoption of e-procurement. Then a research methodology is proposed to study the problem, and this work will contribute to better address the issues faced by suppliers on eprocurement implementations. We plan to apply a quantitative approach to identify the enablers and the barriers that influence Portuguese companies to adopt e-procurement solutions. 183 ISBN: 978-972-8924-89-8 © 2009 IADIS 2. LITERATURE REVIEW 2.1 Supply Chain Management and Collaboration Supply chain management includes the planning and management of all activities included at the sourcing and acquisition process. According to the Council of Supply Chain Management Professionals (CSCMP, 2008), supply chain management encompasses the planning and management of all activities involved in sourcing and procurement, conversion, and all logistics management activities. It also includes coordination and collaboration with channel partners, which can be suppliers, intermediaries, third party service providers, and customers. In essence, supply chain management integrates supply and demand management within and across companies. E-business and e-commerce has a significant impact on level of analysis issues in management research, specifically broadening the perspective to analysis of supply chains and networks (Cromm, 2005). The Web and associated technologies enable collaboration and sharing of information among companies. Collaboration may range from intra-organizational to inter-organizational and across the boundaries of the organization. The procurement is an integral component of an organization’s supplier relationship management strategy, and often is the first major step towards trading partner collaboration (Gilbert, 2000). Collaboration between supply chain members also requires the exchange of sensitive information. Teo & Ranganathan (2004) argued that the heart of B2B e-commerce was in inter-organizational collaboration and it required a fundamental shift in the organizational mindset to collaborate and engage in effective B2B e-commerce. 2.2 E-commerce E-commerce is the process to buy, sell or exchange products or services by the internet. Different models of e-commerce have been presented in order to describe the nature of these transactions. Electronic marketplaces allow collaboration and data sharing within or across industries. It´s possible to classifie the emarketplaces with base on their degree of openess. E-marketplaces with a high degree of openness are those that are publicly accessible to any company. At the other end of the spectrum, e-marketplaces with a low degree of openness are accessible only upon invitation. Based on this distinction, Hoffman, Keedy & Roberts (2002) recognized three main types of e-marketplaces: public e-marketplaces, consortia and private exchanges. Figure 1. Main Types of E-marketplaces The role of procurement and the emerging use of large information systems to conduct e-procurement was analyzed by Hawking and colleagues (Hawking et al, 2004) and presented with the results of a survey of 38 major Australian organizations. The main results show that direct procurement is heavily dependent upon traditional practices while indirect procurement is more likely to use "e" practices. Dedrick (2008) also found that the use of electronic procurement is associated with buying from more suppliers for custom goods but from fewer suppliers for commodity goods. In an efficiently functioning transparent market, few suppliers are sufficient for commodity goods, whereas for custom goods the need for protection from opportunistic vendor leads to the use of more suppliers. 184 IADIS International Conference e-Commerce 2009 2.3 E-procurement Adoption Companies are approaching e-procurement adoption with different strategies. Davila and colleagues (2003) identified two main types of companies. The first type is moving aggressively to adopt e-procurement technologies, frequently experimenting with various solutions. The second type adopts a more conservative strategy by selectively experimenting, typically with one technology. (Davila, Gupta & Palmer, 2003). Also an increasing number of public institutions identified electronic purchasing as a priority to e-government. Many implemented or are implementing e-procurement systems. The adoption of e-procurement in public administration has a huge impact since governments spent large amounts in acquiring materials and services. (Pereira & Alturas, 2007). For the Portuguese government electronic procurement has gained a strategic significance, and some evidences showed that importance. A considerable number of private exchanges and consortia have appeared in the last years. The number of internet transactions has also increased, and the government has deployed some initiatives for public electronic purchasing. The Portuguese government considered e-procurement fundamental for renovation of processes in the public administration (Amaral, Teixeira & Oliveira, 2003). An important step for e-procurement in Portugal was taken by the government with the Decree-Law n. º 104/2002 who allows electronic purchasing by public administration. 2.4 Enablers for Supplier Adoption By understanding the main enablers that influence the supplier adoption of e-procurement, companies can develop strategies to leverage the supplier adoption on an e-procurement implementation. The organization who is implementing an e-procurement solution should assess the impact of the system on suppliers and their technological promptness to implement the system at their end, and should provide the services necessary for the system to succeed. It is also necessary to put together a supplier adoption team, train the suppliers, and get them synchronized with the organization’s implementation (Rajkumar, 2001). Suppliers need to gain conscience of the benefits resulting from their adoption of e-procurement. For suppliers, the adoption of eprocurement may be an opportunity to expand their market. According to Sharifi, Kehoe & Hopkins (2006) they will find e-procurement attractive because they could easily and cost effectively reaches new customers, improving their sales. (Berlak & Weber, 2004). The integration between the buyer and the seller systems allows exchanging information automatically. Therefore, it is possible for the buyer to make an order more quickly. This will also reduce the chance of occurring errors that are common when an order is dependent on paper. (Berlak & Weber, 2004). By linking to a customer directly and collaborating to ensure accurate and on-time delivery provides better service and lower overall procurement costs to the customer, and can result in much more collaborative buyer-seller relationships. (Neef, 2001). Carayannis & Popescu (2005) analyzed and evaluated the electronic procurement projects carried out by European Commission. They concluded that the transparency of EU public procurement market was improved by a systematic use of electronic tendering. The improve on the transparency allow the involved stakeholders to know how the system is intended to work, and all potential suppliers have the same information about procurement opportunities, award criteria, and decisions. In considering how e-procurement will impact buyer-seller relationships Ellram & Zsidisin (2002) argue that close buyer–supplier relationships have a strong positive impact on the adoption of e-procurement. Therefore, while e-procurement may not deliver improved levels of trust, it has been found that eprocurement transactions are more likely to be established first between partners in high trust relationships. In addressing this issue, both Croom (2001) and Kumar & Qian (2006) support the view that increased use of e-procurement and inter-organizational systems enhance opportunities tend to create more effective customer–supplier relationships over time. 2.5 Barriers for Supplier Adoption Cooperation with suppliers also requires them to meet the business criteria that organizations have set to accept them in their networks. Since some of the business models associated with e-procurement technologies clearly envision the use of suppliers with whom the buyer has not previously transacted 185 ISBN: 978-972-8924-89-8 © 2009 IADIS business, companies need to develop mechanisms that provide the buyer with assurances that the supplier meets or exceeds recognizable and industry enforced standards (Davila, Gupta & Palmer, 2003). According to Davila, Gupta & Palmer (2003) providing suppliers with Internet or Intranet access to company internal data, or integrating suppliers applications with company information systems, both key to supply chain management, is still unusual. This observation reinforces the prudence that companies must demonstrate on integrating e-procurement technologies into existing systems and relationships. A study conducted in the Swiss market revealed that the lack of supplier involvement and infrastructure to optimize B2B processes was a hindrance to integrate the B2B solution scenarios. Integration solutions are not always offered appropriate to suppliers and the majority of companies agree that the position of the suppliers is insufficiently considered (Tanner et al, 2008). Buyers are concerned that e-procurement technologies will push prices down to the point where suppliers cannot invest in new technology or product development, upgrade facilities, or add additional productive capacity. Additional price pressures can even push suppliers with a poor understanding of their cost structure out of business (Davila, Gupta & Palmer 2003). Supplier need to know how low they can bid, and still observe an acceptable return. (Moser, 2002). The majority of the companies believe that barriers include insufficient financial support, lack of interoperability and standards with traditional communication. Developing standards and systems for facilitating effective interoperability with traditional communication systems will help the adoption of eprocurement fairly well with minimum investment and changes to the business processes through reengineering (Hawking et al, 2004). Enablers Barriers Market Growth Intranet access Less Operational Cost Integration Transparency Price pressures Longer buyer-seller relationships Lack of standards Table 1. Enablers and Barriers for e-Procurement Adoption by Suppliers 3. RESEARCH QUESTIONS AND METHODOLOGY This research will provide a better understanding of issues affecting the suppliers within an e-procurement implementation. The research questions formulated were based on the enablers and barriers felt by suppliers when confronted with the e-procurement adoption. The following research questions will be answered: • What are the major perceived barriers to the adoption of e-procurement in Portugal by suppliers, and how can they be addressed? • What are the major perceived enablers to the adoption of e-procurement in Portugal by suppliers, and how can companies explore it? This effort started by reviewing the background to the application of e-procurement, which was then followed by various dentitions of e-procurement. Subsequently, we made a review of the literature available on the adoption of e-procurement by suppliers with the objective of developing a theoretical framework for determining the barriers against, enablers and possible solutions for the successful supplier adoption of eprocurement. The questionnaire will be pilot tested by e-procurement consultants and academics, before being sent out. The proposed framework will be validated with the help of empirical data collected from Portuguese companies. Finally, based on the empirical results and analysis, we will develop a framework for the supplier adoption of e-procurement. 4. DISCUSSION Based on the database that we hope to collect, we plan to apply a quantitative approach to identify the enablers and the barriers that influence Portuguese companies to adopt e-procurement solutions. Besides, 186 IADIS International Conference e-Commerce 2009 these empirical evidences could be relevant for managers of companies who seek better understanding and predict the procurement of their products. We hope that companies could leverage their e-procurement implementations by engaging the maximum number of suppliers, and to successfully collaborate on a win to win basis. REFERENCES Amaral, L. A.; Teixeira, C. and Oliveira, J. N., 2003. E-procurement: Uma reflexão sobre a situação actual em Portugal. Minho: APDSI. Berlak, Joachim and Weber, Volker, 2004. How to make e-Procurement viable for SME suppliers. Production Planning & Control, Vol. 15, No. 7, pp. 671-677. Carayannis, Elias G. and Popescu, Denisa, 2005. Profiling a methodology for economic growth and convergence: learning from the EU e-procurement experience for central and eastern European countries. Technovation, Vol. 25, No. 1, pp. 1-14. Croom, Simon R., 2001. Restructuring supply chains through information channel innovation. International Journal of Operations and Production Management, Vol. 21, No. 4, pp. 504-527. Croom, Simon R., 2005. The impact of e-business on supply chain management: An empirical study of key developments. International Journal of Operations & Production Management, Vol. 25, No. 1; pp. 55-73. Crom, Simon R. and Brandon-Jones, Alistair, 2005. Key issues in e-procurement: Procurement implementation and operation in the public sector. Journal of Public Procurement, Vol. 5, No. 3, pp 367-387. CSCMP, 2008. Retrieved 1 25, 2009, from Council of Supply Chain Management Professionals (CSCMP): http://cscmp.org/ Davila, Antonio; Gupta, Mahendra and Palmer, Richard, 2003. Moving procurement systems to the internet: the adoption and use of e-procurement technology models. European Management Journal, Vol. 21, No. 1, pp. 11-23. Dedrick, Jason, 2008. How Does Information Technology Shape Supply-Chain Structure? Evidence on the Number of Suppliers. Journal of Management Information Systems, Vol. 25, No. 2, pp. 41-72. Ellram, Lisa M., and Zsidisin, George A., 2002. Factors that drive purchasing and supply management’s use of information technology. IEEE Transactions on Engineering Management, Vol. 49, No. 3, pp. 269-281. Gilbert, Alorie, 2000. E-procurement: problems behind the promise. InformationWeek, November 20, pp. 48-55. Harris, Lisa, and Dennis, Charles, 2004. Marketing the e-business. USA: Taylor & Francis. Hawking, Paul; Stein, Andrew; Wyld, David C. and Foster, Susan, 2004. E-Procurement: Is the Ugly Duckling Actually a Swan Down Under. Asia Pacific Journal of Marketing and Logistics, Vol. 16, No. 1, pp. 3-26. Hoffman, William; Keedy, Jennifer and Roberts, Karl, 2002. The unexpected return of B2B. McKinsey Quarterly, Vol. 3, pp. 97-105. Kumar, Nanda and Qian, Peng, 2006. Strategic alliances in e-government procurement. International Journal of Electronic Business, Vol. 4, No. 2, pp. 136-145. Moser, Edward P., 2002. E-Procurement--Reverse Auctions and the Supplier's Perspective. Pharmaceutical Technology, Vol. 26, No. 5, pp. 82-85. Neef, Dale, 2001. e-Procurement - From Strategy to Implementation. USA: Prentice Hall. Pereira, Paulo and Alturas, Bráulio, 2007. Factores Críticos da Adesão das PME´S Nacionais, Fornecedoras de Materiais de Escrítorio ao Procedimento Aquisitivo Público em Portugal: Uma Proposta de Investigação. Conferência IADIS Ibero – Americana. Vila Real. Rajkumar, T. M., 2001. E-procurement: Business and Technical Issues. Information Systems Management, Vol. 18, No. 4, pp. 52-60. Sharifi, H., Kehoe, D. and Hopkins, J., 2006. A classification and selection model of e-marketplaces for better alignment of supply chains. Journal of Enterprise Information Management, Vol. 19, No. 5, pp. 483-503. Tanner, Christian; Wolfle, Ralf; Schubert, Petra and Quade, Michael, 2008. Current Trends and Challenges in Electronic Procurement: An Empirical Study. Electronic Markets, Vol 18, No. 1, pp. 6-18. Teo, Thompson S. H. and Ranganathan, C., 2004. Adopters and non-adopters of business-to-business electronic commerce in Singapore. Information & Management, Vol. 42, No. 1, pp. 89-102. 187 ISBN: 978-972-8924-89-8 © 2009 IADIS MEASURING VIRTUAL KNOWLEDGE MANAGEMENT IMPACT IN FIRM’S PERFORMANCE Flávio Gomes Borges Tiago, Maria Teresa Borges Tiago, João Pedro Almeida Couto Universidade dos Açores ABSTRACT Knowledge Management (KM) is one tool that seams to gain a more relevant role, especially as managing knowledge has become increasingly important to all companies. Appropriate KM practices within organisations can be seen as one of the prerequisites to the enhancement of continuous performance improvement in the interne-based context. Thus, our aim is to develop a conceptual framework related to KM practices in a virtual context and to identify the nature of the relationship existing in those knowledge-driven elements and performance achievements. This paper aims to bridge the gap between the KM and e-business performance-related literatures by establishing a model tested in European companies using a structural equation modelling analysis. The results show that KM has a positive impact on the maximization of e-business performance and that some elements individually have a positive influence on e-business performance. The present study advances knowledge on the nature of the relative importance of different components of Internet-based KM as drivers of e-business performance and reinforces its importance as an integrated e-business tool. KEYWORDS Knowledge Management, e-Business, Performance. 1. INTRODUCTION In today's digital economy, rapid access to knowledge is critical to the success of many organizations (Liao, 2003). One of the major challenges that firms face is managing competitive advantage through the development of strong relationships with all stakeholders. In this context, Knowledge Management (KM) becomes an important part of the global solution. However, as noted by Takahashi and Vandenbrink (2004) and Zhang and Zhao (2006), KM needs to be regarded as more than simple information gathering in order to take advantage of its competitive potential. Despite the academic research and organizational practices developed around this concept, there is still a lack of conceptual basis necessary to develop the measures of KM contribution in business success, especially regarding its contribution to Internet-based environments. The objective of this paper is to gain a clearer understanding of the fundamental issues related to this topic. In this line of research, the present paper discusses the results of an exploratory survey conducted among a large sample of European companies. Using a structural equation analysis, we explore the relationship between ebusiness performance and KM initiatives, trying to identify the main drivers of virtual KM. 2. THE VIRTUAL KNOWLEDGE MANAGEMENT Many claim that knowledge is a major factor driving business-level capabilities and an important source of competitive advantage (Nonaka & Takeuchi, 1995). Awad and Ghaziri (2004) stated that information and knowledge are critical to companies’ performance. However, these authors suggested that capturing and transferring best practices is not enough to achieve success, especially in an Internet-based context. The expansion of Internet and e-commerce technology allows firms to establish new forms of creation knowledge, and provides them opportunities to improve their capability to manage and use knowledge (Siau, 2000). Through the Internet, vast amounts of information concerning customers, suppliers, markets, and supply chains can be effortlessly gathered, while information about company processes, products, and services can be easily disseminated to the public. 188 IADIS International Conference e-Commerce 2009 Takahashi and Vandenbrink (2004) suggested that the problem facing top decision-makers in the ubiquitous information society will be how best to organize the knowledge cycle. One of the challenges is to share the knowledge with inside entities who value it, and to do so organizations must create and deploy knowledge management systems (KMS). KM is one of the leading strategic areas being explored and adopted by companies (Schwartz et al., 2000; Grossman, 2006), especially by those who have invested in the Internet as a new channel and marketplace. According to Stojanovic and Handschuh (2002), the main function of a KM system is to capture and disseminate new sources of information. From this point of view, the Internet is a font of information. By using the Internet, companies implement a knowledge-acquisition and knowledge-sharing system, one that meets the requirements and specifications of unique and complex systems. It will match customer requirements to product characteristics (Ratchev et al., 2003) and allow the acquisition and maintaining of competitive advantages. Furthermore, in this digital society, corporations need to adapt both knowledge management systems and business strategy in order to use digital information effectively and to take advantage of Internet possibilities (Takahashi & Vandenbrink, 2004). Like many other information system implementations, KM is strongly linked in the literature to a sales and marketing perspective (Zhang & Zhao, 2006). For this research, we will consider KM as a combination of marketing tactics, knowledge-sharing, methods and technology. It can be used to gain and maintain competitive advantages in a global marketplace such as the Internet and simultaneously cut down organizational layers. As Malhotra (2000) suggested, the traditional KM model emphasizes convergence and compliance to achieved pre-specified organizational goals. On the other hand and according to several authors, virtual KM emerged from the Internet, and web technology facilities are used to implement KM solutions. Nevertheless, the concept of use of information technology as the key enabler of KM is not a new idea. From the literature review performed, we consider virtual KM as an Internet-based business strategy integrating every area that touches the data gathering. These areas include sales and support services, the overall consideration of enhancing performance of people and processes with major contributions from new electronic technology (Internet, email, chat rooms, e-forums), and data transformation into information, i.e., extranet and other internal process and knowledge-sharing (intranet, extranet, LAN, WAN, VPN). In this context, online companies are embracing knowledge management as a major element of corporate strategy. Online technological applications allow a rapid and low-cost access to data, faster and easier processing of the information and, above all, a greater level of knowledge sharing. However, the adoption of KM systems by online organisations implies a complex restructuring of all organisational elements and processes, in order to achieve the competitive advantages through the use of virtual KM systems. The virtual KM can be define has the incorporation of online technologies in the cycle of knowledge in order to enhance the KM processes. 3. EVALUATION FRAMEWORK AND HYPOTHESES Following the literature reviewed in the previous section, we developed a research model. It proposes virtual KM that will be positively associated with a set of intermediate outcomes that we call “KM practices”, and will be positively associated with online organizational performance. For that purpose we use a structural equation model with latent variables. This model consists of two sub-models: the measurement model and the structural equation model. The primary research questions to consider are these: What is the degree to which an organization engages virtual KM — in particular, technological KM practices — has a positive impact in online organizational performance? And is virtual KM, in turn, positively related to online organizational performance? Besides measuring the convergence of a technological approach with a business value approach, our aim is to discover the direct nature of the relationship between KM practices and online organizational performance. The validation of the measurement model is done by using Confirmatory Factor Analysis (CFA). We will see later that the observable variables (indicators) we selected are measures of three latent variables (factors). We assume that these three KM practice factors each have a direct effect on the virtual KM and upon online corporation performance. Therefore, we assume that the online corporation performance is explained not 189 ISBN: 978-972-8924-89-8 © 2009 IADIS only by the virtual KM, but also by a general KM practices factor that is concerned with the gathering of data, information process and knowledge-sharing. Therefore, it is postulated that the considered indicators measure three different and positively correlated latent variables or factors (hypothesis H1). Each factor is supposed to contribute directly to the determination of the online corporation’s performance (hypotheses H2 and H3). Besides these direct effects, it is also assumed that there is an indirect effect via virtual KM (H4). In sum, the four research hypotheses are the following: H1: The indicators considered define three positively correlated factors; H2: The KM practices factor positively and significantly determines the online corporation performance; H3: The factor concerned with KM practices positively and significantly determines the virtual KM; H4: The KM practices influence online corporation performance through virtual KM application. Awad and Ghaziri (2004) pointed out that KM awareness benefits the entire organization and that it relies on developing a KM environment inside and outside the firm — one that permits a generation of new knowledge, i.e. the transfer of existing knowledge and its application to new products, services and process. Davenport and Prussak (1998) considers that KM focuses on processes and mechanisms for locating and sharing knowledge possessed by an organization or its external stakeholders. Based on this, we define KM practices as the group of technological efforts carried out by the organization in three different dimensions: data gathering, information process and knowledge-sharing. In total, we identified twelve KM practices. Each has been suggested elsewhere as being important for effective virtual KM (Gold et al, 2001; Malhotra, 2000; Awad and Ghaziri, 2004; Schwartz et al., 2000; Tiago et al, 2007; among others). In Internet-based practices, most traditional financial and accounting methods of evaluation are not suitable as the only forms of performance measurement. This is due to the fact that there are some intangible, indirect and even strategic benefits that need to be considered. From the literature review, it is found that KM has been linked positively to non-financial performance measures such as quality (Mukherjee et al., 1998; Tiago et al., 2007), innovation (Francisco & Guadamillas, 2002), productivity (Lapre & Wassenhove, 2001), and sales (Tiago et al., 2007). In fact, the expected results are that KM simultaneously influences many different aspects of organizational performance. The work of Gold et al. (2001) presents a combination of two dimensions as enablers of effective performance improvements: knowledge infrastructure and knowledge-processing capacity. Other frameworks have been presented, but the specific interface between virtual KM and e-business has not been addressed from the organisational point of view. So we will follow in the last authors’ steps, using as performance measures elements of both infrastructure and processing dimensions. By identifying KM practices as antecedents to virtual KM and online organizational performance, we attempted to include factors that have been previously tested by others authors (see for example, Gold et al., 2001). 4. METHODOLOGY AND RESULTS The data used to test our research model comes from the e-Business W@tch annual survey (2005). This data was collected in a large survey about e-business in European enterprises. Considering that this study examines the status of adaptation of virtual KM by companies, the original sample was limited to firms having e-business activities and companies adopting KM practices. So, our work sample of 5.216 cases constitutes a heterogeneous sample of companies in terms of industries, fields, size, business model and country. The data covers 7 European countries (Czech Republic, France, Germany, Italy, Poland, Spain and the U.K.). Distribution of firm size, measured by the number of employees, shows that almost half of the firms are micro- and small-size firms (around 50.7%). The industry distribution of the responding sample is approximately similar to the original sample. The two less heavily represented sectors in the sample are the aerospace industry and manufacture of pharmaceuticals, with 3.1% and 10.2% respectively, closely followed by all the others. The model was estimated by the Maximum Likelihood method in the AMOS package. The model goodness of fit may be considered acceptable according to the values of some goodness-of-fit index, although the chi-square test statistic (χ2 = 626,4; df =117; p-value = 0,000) is significant, implying a bad fit. However, as is well known, this test has serious limitations — namely its dependence on the sample size and on the number of indicators. In general, for large sample sizes the chi-square statistic is significant, and in the 190 IADIS International Conference e-Commerce 2009 present case the sample size is very large (n = 5,216). For that reason, it is usual to evaluate the goodness of the fit by a set of index, also presented in Figure 1. After global model fit has been assessed, the numerical results were evaluated in order to test their support of the research question. The numerical results can be obtained directly from the path coefficients of the structural model. We refer to standardised coefficients that account for scale effects and serve as indicators of the relative importance of the variables. The measures for global model fit included in Figure 1 suggest that our model fits the underlying data well. All the paths were statistically significant. The three dimensions used to compose the KM practices are all significant and highly correlated to the KM practices construct. As a result, hypothesis H1 is not rejected. Nevertheless, a reference needs to be made regarding the relative lower value achieved in terms of knowledge-sharing. The results show that KM practices competencies explain only 11 percent of the variance in online corporations’ performance. Thus, this finding gives no empirical support to the concept that online performance can be improved by the use of the three basic components of KM traditional practices: data gathering; information process and knowledge-sharing. With this consideration in mind, hypothesis H2 is rejected. The results also show that virtual KM explains 51 percent of an online corporation’s performance, implying that our hypothesis H3 is not rejected. The data gathering, information processing and knowledgesharing combined are not significantly important for the direct determination of an online corporation’s performance. However, these items have an indirect effect on the performance via their positive influence on the virtual KM. So, hypothesis H4 is not rejected. KM practices and virtual KM are only part of the equation; the construct of online corporation performance must also be measured. All of the non-financial factors used show a positive and significant relationship. This provides empirical support for the theoretical views that state that online performance needs to be measured using new criteria, and not exclusively finance-based criteria. Data Gathering Information Process *** 1,00 1,00 0,11 # *** Online Corporations performance KM practices *** 0,51 # 0,55 Knowledge sharing 0,31 *** Key for significance measures: *: α >0.10 **: α >0.05 ***: α >0.01 #: for model identifiably, this path coefficient was set to 1 in the unstandardized case. Virtual KM Measure Value Suggested RMSEA 0.033 < 0.05 NFI 0.900 >0.9 IFI 0.913 >0.9 CFI 0.913 >0.9 Figure 1. Structural Equation Model and Estimation Results 5. DISCUSSION AND CONCLUSIONS Knowledge Management has presented several difficulties in the traditional IT environment, basically related to the constrained form of sharing the knowledge. In the present ubiquitous information context, KM seems to be an easier and promising tool, especially when used in its global version. As the literature review showed, there have been only a few works examining KM practices and virtual KM contributions to online performance from a corporate perspective. Moreover, the majority of these works were confined to specific industries and confined to small data samples. The goal of the current study was therefore to answer the following questions: What is the degree to which an organization engages virtual KM — in particular, technological KM practices — has a positive impact in online organizational performance? And is virtual KM, in turn, positively related to online organizational performance? With this study, we attempt to contribute to bridging the existing research gaps. We do so by presenting results from an empirical investigation based on a cross-industry survey, which covers seven European countries. The findings shown above, as reported by respondents in the case companies, demonstrate the kinds of applications they really need or value, how KM practices are used and valued, and the ways in which virtual KM can help to achieve higher levels of online performance, considering a new set of non-financials measurements. Considering the results, we can find evidence to confirm most of the hypotheses that we formulated regarding the impact of virtual KM in online corporations’ performance. First, the data supports our 191 ISBN: 978-972-8924-89-8 © 2009 IADIS conceptualisation for the KM practices construct: data gathering, information process and knowledgesharing. Within this, all elements have a positive impact on the maximisation of KM practices. Secondly, the findings allow us to conclude that virtual KM has a positive impact on online performance, which was expected considering the existing literature on this matter. According to these results, the concept of virtual KM as an important e-business tool is reinforced. Thus, the relationship between virtual KM and online performance follows the positive relationship found in some earlier studies. One of the managerial contributions of this work is the discovery that managers should consider the use of virtual KM to improve everyday online processes — and should also be aware that the simple use of the KM practices is not enough to achieved higher performance levels. However, a cost–benefit analysis should be made to assess the return on the investments made in KM, since we only considered the upside of this initiative. Until KM becomes an ingrained and standard tool of e-business, the need to define measurement criteria will continue in order to support the corporate implementation and maintenance of such systems. Further work is clearly needed to examine the interaction between virtual KM and online performance over time or in small sets of the sample. Doing so would allow us to find out if the relationship is equally strong in all countries and which contextual factors affect this relationship. This research produces some useful insights, leaving still a number of issues for future research. One of these issues is related to the possibility of extending the study in order to consider the impact of other elements of virtual KM, such as technological readiness and management support. Similarly, this study could be expanded through the application of a panel data methodology that would determine the evolution of virtual KM contribution to online performance among European companies. REFERENCES Awad, E. and Ghaziri, H. (2004) Knowledge Management. New Jersey: Pearson Education. Davenport, T. and Prussak, L. (1998) Working knowledge: how organizations manage what they know. Boston: Harvard Business School Press. Francisco, J. and Guadamillas, F. (2002), A case study on the implementation of a knowledge management strategy oriented to innovation. Knowledge and Process Management. 9(3): p. 162-171. Gold, A., Malhotra, A. and Segars, A. (2001) Knowledge management: An organizational capabilities perspective. Journal of Management Information Systems. 18(1), 185-214. Grossman, M. (2006) An overview of knowledge management assessment approaches, Journal of American Academy of Business. 8(2), 242-247. Lapre, M. and Wassenhove, L. (2001) Creating and Transferring Knowledge for Productivity Improvement in Factories. Management Science. 47(10). Liao, S. H. (2003) Knowledge management technologies and applications - literature review from 1995 to 2002. Expert Systems with Applications. 25 (2), 155-164. Malhotra, Y. (2000); Knowledge Management for E-Business Performance: Advancing Information Strategy to ‘Internet Time’, Information Strategy: The Executive's Journal. 16(4), 5-16. Mukherjee, A., Lapre,M. and Wassenhove, L. (1998) Knowledge Driven Quality Improvement. Management Science. 44(11), S35-S49. Nonaka, I. and H. Takeushi (1995) The Knowledge-Creating Company, Oxford: Oxford University Press. Ratchev S., Urwin E., Muller D., Pawar K.S. and Moulek I. (2003) Knowledge based requirement engineering for oneof-a-kind complex systems. Knowledge-Based Systems. 16 (1),1-5. Schwartz D. G., Divitini, M. And Brasethvik, T. (2000) Internet-Based Organizational Memory and Knowledge Managemen. Hershey: Idea Group Publishing. Siau, K. (2000) Knowledge discovery as an aid to organizational creativity. Journal of Creative Behavior. 34(4), 248– 258. Stojanovic, L.Stojanovic, N. and Handschuh, S. (2002) Evolution of the Metadata in the Ontology-based Knowledge Management Systems. German Workshop on Experience Management. Takahashi, T. and Vandenbrink, D. (2004). Formative knowledge: from knowledge dichotomy to knowledge geography – knowledge management transform by the ubiquitous information society. Journal of Knowledge Management. 8(1), 64. Tiago, M., Couto, J., Tiago, F. and Vieira, A. (2007) Knowledge management :An overview of European reality. Management Research News. 30(2), 100-114. Zhang, D. and Zhao L. (2006) Knowledge Management in Organizations. Journal of Database Management. 17(1), i-viii 192 IADIS International Conference e-Commerce 2009 MODERN ARCHITECTURAL REASONING FOR COMPLEX WEB COMMERCE APPLICATIONS Thomas Lehrner Culturall Handelsges.m.bH. 1040 Vienna, Austria Birgit Pohn University of Applied Sciences Technikum Wien 1200 Vienna, Austria Markus Schranz Vienna University of Technology 1040 Vienna, Austria ABSTRACT Within the Information Age, commerce has found a channel of distribution which has gained a new boost through the Web 2.0 phenomenon. Rich User Applications coupled with the spread of the Internet and WWW has spanned a global mesh of communication and online services which does not depend on any temporal or local factors. Simplicity and ubiquitousness of modern web applications have become the pedestals of spreading e-business systems all over the world. This ease of access and use for customers poses a challenge to service-providers. Designing and developing web 2.0 software depends on the one hand on the discussion of architectural concepts for distributed systems and on the other hand on appropriate frameworks which allow building up complex web-applications e.g. the well-known webapplication-framework MASON. This article deals with topics like concepts of implementing business processes as webservices, desktop-like user interfaces and persistent mass transactions in comparison to each other. We propose an up-todate object-oriented approach to the development of modern e-commerce applications based on MASON and provide profound theoretical and practical results based on an industry e-commerce-solution. KEYWORDS e-Commerce Systems, Web Publishing, Web Application Architectures, Web 2.0, Mason. 1. INTRODUCTION Through the start of the WWW 15 years ago, the world has undergone one of its most drastic steps towards globalisation. As a medium that reaches the whole world, the Web is the logical continuation for setting up business-processes. On its way from static information sites in the mid 90s to user created contents, sharing the knowledge of the masses and providing easy access to the long tail in applications of the Web 2.0 (O’Reilly, 2005) the Web has traversed through various technical stages. Multiple approaches, concepts and architectural designs have been developed in order to create and maintain user-oriented comfort and technical simplicity for viewers and end customers while accessing extensive Web services. Technicians, IT and content architects, marketing experts and sales professionals invented Web services and applications in the area of information supply, news networks (Dustdar, 2006), content management, cultural resources (Schranz, 2005), etc. With increasing volume and complexity of modern Web applications the technical challenges demanded sophisticated software engineering skills (Ghezzi, 2002) and highly developed tools, evaluated by research methodologies (Kirda, 2001; Kappel, 2006) for providing and sustaining successful services. Especially the ambitious entrance of commerce to the Web has introduces requirements to multiple disciplines such as security, performance, service performance and software architecture. In this paper we discuss technologies and approaches to copy with the challenges of complex Web services and propose appropriate architectural elements to implement and maintain modern commercial information applications on the WWW. 193 ISBN: 978-972-8924-89-8 © 2009 IADIS The article is structured as follows: in section 2 we refer to basic definitions, widely applied and researched technologies and related concepts. Section 3 focuses on Web Commerce and the challenges in combining traditional commercial transactions with modern distributed software architectures in high volume contexts. Section 4 proposes a component-based object-oriented approach to complex web architectures and section 5 explains our research experiences based on industry applications recently implemented following our approach. The summary provides our conclusion and future steps to optimize future assembling of appropriate architecture components for successful web commerce applications. 2. BASIC TECHNOLOGIES AND ARCHITECTURE UTILIZED IN WEB APPLICATION ENGINEERING In order to provide and maintain (commercial) web applications, a set of basic technologies are utilized on various platforms, thus creating a manageable family of web architecture components. On top of the basic technologies and concepts modern Web applications require data persistency, (a)synchronous information exchange (concerning rich user experience) and modern communication and service decomposition approaches for flexible and manageable architectures. 2.1 The Hibernate Persistency Framework Data persistency frameworks such as the open-source Java Framework Hibernate (Elliott, 2008) have been developed to reach data persistence by storing and reloading complete objects including their dependencies in relational databases by mapping them to database-tables. Working with Hibernate facilitates accessing data from any database since there's no need to build up any database-connections or set up SQL-statements to guarantee data persistency whereby development time is reduced. Diag. 1. Session-per-conversation-pattern (a) Vs. Session-per-request Pattern (b) An essential benefit of Hibernate is the architecture of “communication”. Within one conversation (a complete user-dialog, e.g. buying a ticket) the Hibernate Session holds any data, state information and manages several database transactions to broaden the scope of the actual session for further processing simulating a continuous database-connection1. Basically two communication approaches are distinguished in Hibernate: transaction-spanning-conversations and the session-per request-pattern (see diagram 1). 2.2 AJAX – Asynchronous JavaScript and XML AJAX enables asynchronous communication between client and web server. In contrast to conventional webapplications, modern web-applications using AJAX can exchange data independently from user-generated HTTP-requests. The main idea of using AJAX is providing faster and more interactive services which behave almost like desktop-programs and don't keep the user waiting. AJAX allows loading information belatedly and reacting to user-actions and behaviour autonomously. 2.3 SOA – Service Oriented Architecture A principle concept of modern application architecture is the Service-Oriented-Architecture (SOA). Core of this approach is structuring services, which can be derived from splitting up complete business processes into 1 194 Hibernate source at www.hibernate.org/42.html, accessed on July 15, 2008 IADIS International Conference e-Commerce 2009 logical atomic entities, and provide them as independent modules within a distributed system for reuse. Together with isolated use of a service it is also possible to combine different modules for realising a certain process – the coordination of those elements is called service-orchestration. Service-orientation means customer-orientation as well – by providing single services and tailored orchestrations delivered data can be exactly adapted to specific demands. 3. TRADITIONAL/MODERN CONCEPTS FOR WEB APPLICATIONS Ever since the Internet pioneers like Tim Berners-Lee, Marc Andreesen and many others have provided the necessary basics to web communication and interactive distributed applications, research and the information industry have been utilizing the merits of their achievements. The technical basics in HTTP as a protocol, HTML as simple yet powerful markup language and URL as a unique addressing recommendation are standardized by the W3C. Due to the simplicity of the protocol, the number of web applications provided have raised from a few hundreds in the early 90ies to more than 170 million sites2 in June 2008. From early static information services the application categories (cf. Kappel, 2006) provided at the Web have expanded to interactive, transaction oriented, user-centered and ubiquitous Web products with similar user experience as locally installed software and transporting semantically related information networks to the target users. 3.1 Challenges in Combining Commerce and Web Applications The analogy to face-to-face shopping misleads the users to the assumption that a commercial web service is solely provided for their comfort. Commercial terms such as “one-to-one marketing” emphasize this illusion, but commercial providers in contrast create maximum financial benefits from top frequencies and multiple parallel requests, which have to be managed by modern service architectures, to the product servers. Modern Web applications support database concepts such as atomicity, concurrency, isolation and durability for commercial transactions. In contrast to locally installed applications, where direct and permanent access to the main database host can be provided and guaranteed, web applications are immanently distributed. The client/server architecture, which came with appealing simplicity to spread the network all over the world, has tremendous drawbacks, when it comes to multi-request transactions. Due to the distributed hosts involved in the transaction there is at no time a guarantee, that the conversation can be continued. A disconnected might happen at any time during the commercial transaction, leaving the commercial application in a problematic state. 3.2 Persistency and Commerce To manage the challenges of commercial transactions in a distributed service environment, software and Web architects have invented different approaches and technical concepts. Similar to the solutions provided in modern database technology (cf. the two-phase commit protocols or the ACID-attributes of local transact ions), distributed conversations have to overcome network insufficiencies with reliably persistency models. For such distributed data and transaction management various persistency frameworks have evolved: e.g. the Hibernate-Framework for Java Applications (section 2.3.1.) and the Apache::Session Module for LAMPP (Linux/Apache/MySQL/Perl/PHP) architectures. 3.3 Software Architectures for Modern Web Commerce Applications Since modern Web applications have soon outgrown the stage of single page monolithic software scripts, the creation and maintenance of current services need to be managed by appropriate processes and tools (cf. Kirda, 2001). Independent of concrete middleware, application services or programming languages, most web applications share commonly identifiable software and architecture patterns. 2 Source: news.netcraft.com/archives/web_server_survey.html visited at July 15, 2008 195 ISBN: 978-972-8924-89-8 © 2009 IADIS In order to guarantee maintainability and quality, an appropriate distribution of concerns have to be established. An approach in software development, recently adapted to the needs of Web applications (cf. Ping, 2003) is the MVC (model-view-controller) pattern, which forces the separation of business logic, user interface or represented data. Modern frameworks following the MVC pattern include Ruby on Rails, Jifty, MasonX::MiniMVC and many others. These frameworks usually simplify the web application development by providing automatically generated software components within the MVC pattern, to be extended and finished by the programmers. 4. ARCHITECTURE ELEMENTS FROM A COMPONENT-BASED OBJECT-ORIENTED APPROACH TO WEB ENGINEERING Based on modern concepts in designing and developing complex Web commerce applications we have identified and compared software components that are appropriate to creating modern and successful services. Our selection is based on the web application service framework Mason (Rolsky, 2002). Mason has been designed to build, organize and maintain large dynamic web sites – all based on combining its atomic units: components. It encourages thinking in structural terms and architectural dimensions rather than procedural scripts or modules. We describe selected architecture elements and their representation in Mason in the following: 4.1 Component Inheritance and Wrapping Chain Components are related to each other in arbitrary hierarchical order. Mason components can inherit from other components, like classes and objects in an object-oriented system. Normally, each component will inherit form a single component called the autohandler, or it may specify a particular parent or declare no parent. The default parent for an autohandler is an autohandler in the upper directory within the scope of the document root of the web server. The autohandler implements general behaviour for components, like the web site structure. Mason processes a request by building a wrapping chain and executing each component in that chain from the topmost parent to the bottommost child. The requested component is always at the bottom of the chain. Inside the wrapping chain all components could call other components or methods thereof. After each execution of a component, Mason could filter the output. This feature offers additionally to the wrapping chain construct a component dependent output adaptation. 4.2 Persistent Session and Data Handling Persistent session handling is required to maintain user specific data spanning several requests. This behaviour is not implemented directly in Mason, thus giving developers total control over session handling, such as session id communication or any arbitrary data structure identified uniquely by the session id (based on cookies or URLs). The storage and locking mechanism to guarantee mutual exclusion on the access to the session data can be chosen to support consistency and security of the data in the most performing way. Complex web applications often greedily require resource intense calculations and database queries, where results typically remain unchanged for a specified time period. To save system resources, entire components may be cached for that reason by the Mason data caching interface to improve performance. Each component has its own persistent data cache and each cached value is set to expire after a specified time period. 4.3 Persistent Database Handles Utilizing the most popular web server Apache, and one of its most powerful extension mod_perl, Mason is employed as modern component in a powerful architecture. CGI is the most widely used protocol for building database interfaces, implemented in Apache's mod_cgi and its equivalents. The main limitation is the lack of persistent connections to the database. Normally, for 196 IADIS International Conference e-Commerce 2009 every HTTP request, a process has to connect to the database, and on request completion the connection will be closed. Depending on the database, the time to instantiate a connection may vary significantly. Apache::DBI (Descartes, 2000) was written for this purpose. Used with mod_perl, the DB connection persists for the entire lifetime of an apache process. That is possible, because a child process of apache does not quit when a request has been processed. When mod_perl code needs to use the DB, Apache::DBI immediately provides a valid connection since it is cached in the apache process - if it was already established earlier - and the Mason resp. Perl code can be processed further. 4.4 Transactions per Request Model A single session typically has a bigger scope than a single database transaction. This implementation pattern includes all transaction benefits and might span several database transactions, but without excessing of the request boundaries. The Transactions per request model shows the logical idea of a conversation and how it can be setup to map a process. The session is held persistently in disconnected state and does not require a permanent open database connection, during user inactivity. The Apache-Mason session maintains all userspecific data between requests. Diagram 1 in section 2.3 shows the idea of this pattern. 4.5 Rich Internet Applications with AJAX Rich Internet Applications (RIAs) are web applications that show features and functionality of traditional desktop applications. Typically, RIAs run in a web browser and are platform-independent. A major difference between traditional web applications and Web 2.0 is performance - realized by outsourcing parts of the application to the client. The consequence is a major saving of process resources and bandwidth, because only requested components have to be computed and therefore less bandwidth is used to transfer them, which increases the reaction time of a web application. 4.6 Integrating Mason Features by the MVC Pattern The model-view-controller (MVC) architectural pattern, as described in section 3.4, isolates business logic from user interface aspects. Our approach prescribes specific coding standards that enforce the MVC pattern. Consequently we separate software components on order to more easily modify complex web applications, so that one layer does not effect the other. Mason makes no attempt to enforce any sort of discipline in the design. Instead, this is the responsibility of the developers and application designer. Mason is at its core a text templating tool, which is realized with the wrapping chain(cf. section 4.2. All chain components can call all outside components. This functionality perfectly represents the view logic of the MVC architecture pattern. 5. ARCHITECTURAL RESULTS AND EXPERIENCE IN COMMERCIAL SERVICES In combining the architecture components and concepts described above, we identified and implemented a set of appropriate concepts, software patterns, and tools to comfortably and sustainably realize an ecommerce solution. Exactly the above-mentioned architectural elements have been supporting our research and development work to design, monitor and maintain a successful commercial application: 5.1 Culturall Opera & Theatre E-Ticketing Service Most recent deployment of the research results is available at the commercial website of the Culturall HandelsgesmbH, Vienna, Austria. Ticket sales are managed directly via box offices at operas, theatres, museums, ticket agencies and via the Web. Culturall faces a major difference to standard internet sales, like Amazon or other Internet stores, since the Culturall distributed system requires real-time information of tickets. A seat, which is selected by a customer in the internet, cannot be sold over another distribution channel at the same time (mutual exclusion). 197 ISBN: 978-972-8924-89-8 © 2009 IADIS These real-time requirements have to be implemented in a performing and resource saving way. The Web platform is developed with Mason web application framework, thus guaranteeing these features. Mason components are easily reused and caching minimizes database access and reduces response time. Additionally, AJAX is used to improve user experience, interactivity and “invisible” server requests e.g. loading real-time parameters for performance schedules. 6. CONCLUSION In this paper we have discussed basic concepts of Web technology and provided services and combined modern architectural elements to a successful set of concepts and tools in order to design, implement and maintain modern web commerce applications. MVC-pattern based architectures and SW-concepts integrated in the Mason Web application server framework are the core of our architecture. The successful implementation of the Culturall E-Ticketing service underlines the viability of our approach. Optimization in the utilization of the MVC pattern and improvements in the testability of asynchronously generated AJAX site components are important next steps in our research work. ACKNOWLEDGEMENT This research work has been conducted with support of Culturall Handelsges.m.b.H., Vienna, Austria. The authors express their gratitude to Dr. Helmut Rainel for providing access to software components and commercial contents that allowed thorough research on the discussed field studies. REFERENCES Bekman, S. & Cholet, E. 2003. Practical mod_perl. UK: O'Reilly Descartes, A. & Bunce, T. 2000. Programming the Perl DBI. UK: O’Reilly Dustdar, S. & Schranz, M.W. 2006. Multimedia News Systems, in Encyclopedia on Multimedia. USA: Springer, pp435441 Elliott, J., Fowler, R., & O'Brien, T. 2008. Harnessing Hibernate. UK: O'Reilly Ghezzi, C., Jazayeri, M., & Mandrioli, D. 2002. Fundamentals of Software Engineering. USA: Prentical Hall Kappel, G., Pröll, B., Reich, S. 2006. Web Engineering. The Discipline of Systematic Development of Web Applications. UK: Wiley & Sons Kirda, E., Jazajeri, M., Kehrer, C. & Schranz, M.W. 2001. Experiences in Engineering Complex Web Services. IEEE Multimedia V8(1), pp58-65 O'Reilly, T. 2005. What Is Web 2.0?. O'Reilly Network, http://www.oreillynet.com/pub/a/oreilly/tim/ news/2005/09/30/what-is-web-20.html, last download 14-jul-2008 Ping, Y., Kontogiannis, K., and Lau, T. 2003. Transforming Legacy Web Applications to the MVC Architecture. Proceedings of the Eleventh Annual International Workshop on Software Technology and Engineering Practice,STEP. Amsterdam, Netherlands, pp133-142 Pohn, B. 2007. Buzzword Web 2.0 -Ideen und Technologien der zweiten Version des WorldWideWeb. University of Applied Sciences Burgenland, Master Thesis Rolsky, D. & Williams, K. 2002. Embedding Perl in HTML with Mason. UK: O’Reilly Schranz, M.W. 2007. Cultural Content Management at a New Level: Publishing Theater and Opera Details by Means of Open Technologies from the Web 2.0. Proceedings of the ICCC 11th International Conference on Electronic Publishing – ELPUB 2007. Vienna, Austria, pp 444-450 Schranz, M.W. 2005. Semantically connecting Business News in Europe. Proceedings of the SEMANTICS 2005 Conference. Vienna, Austria, pp 48-56 Thomas, D., Hansson, D., Breedt, L, & Clark, M., Davidson, J.D., Getland, J., & Schwarz, ª 2006. Agile Web Development with Rails. USA: Pragmatic Pookshelf Verlag. 198 IADIS International Conference e-Commerce 2009 A NEGOTIATION MECHANISM USING ARGUMENTBASED METHODS AND PERIPHERAL ISSUES IN MULTIAGENT SYSTEMS WITH INCOMPLETE INFORMATION Azita Darooei Computer Department , Faculty of Engineering , University of Sheikh-bahaee , Isfahan , Iran Mohammad-Reza Khayyambashi Computer Department , Faculty of Engineering , University of Isfahan , Isfahan , Iran ABSTRACT In this paper a model of negotiation in multi-agent systems is represented. Overall aim of the model is to increase the speed of agents’ agreements in a way that causes an improvement in obtained utility in some cases (in comparison with ordinary methods). Based on the proposed method, each agent as one side of a negotiation, can use its peripheral issues (issues which are available in its optimal agenda, but are not discussed in the other agents’ optimal agendas) to reach agreement in a shorter time. In this model which is a competitive-cooperative one indeed, each agent accepts to concede a privilege to its competitor (cooperative), and in return receive another privilege from the competitor (competitive). This operation is done in a way that not only brings about no utility loss, but also increases utility in some cases. The reason is that the probable loss will be compensated through the received privilege. We aim to provide these privileges by using peripheral issues and argument-based methods. KEYWORDS Negotiation , e-commerce , agent , argument-based method , peripheral issues , incomplete information. 1. INTRODUCTION Electronic commerce is one of the most important ways of trading especially because of its easy, rapid, and global nature. Briefly, e-commerce consists of the buying and selling of products or services over electronic systems such as the Internet and other computer networks [11]. Nowadays automation of electronic trades is one of the most vital issues in e-commerce areas, and intelligent agents are utilized for gaining this purpose [7]. Negotiation is a way of interaction between these agents in e-commerce contexts and multi-agent systems [1].The main reason of this matter could be found in differences between agents preferences and their attempt to reach faster agreements [2,10]. Obviously, the source of these differences stem from selfinterested nature of agents and their final intention of gaining the maximum utility for their selves. Negotiation is used as a means of regulating these differences, and afterwards the agents can operate together. Another problem is the existence of time limitations and predefined deadlines for negotiations. Since, two sides of a trade try their best to reach an agreement before negotiation time ends. Achieving this goal in multi-agent systems with complete knowledge (which each side of negotiation knows about the limitations or preferences of the other side) is so easy. However, despite all the efforts in this domain, achieving this goal in multi-agent systems with incomplete knowledge still is a big problem. On the other hand , as the time passes usually agent’s utility decreases and in fact its curve has a descending shape. We call this utility decrease “concession rate”. Concession rate is another reason of agents’ motivation toward quicker agreements [8]. Based on concession rate principal , if agent a gives offer Si at time t and offer Si+1 at time t+1 , with maximum probability the utility gained through offer Si is more than Si+1 (e.g. if U is utility function for agent a: Ua(Si)≥Ua(Si+1) ). Another subject that has fundamental effects on negotiation outcomes is method and the model of 199 ISBN: 978-972-8924-89-8 © 2009 IADIS negotiation between negotiators. As we know, different models of negotiation are represented but each one suffers from some limitations. One class of negotiation models are argument-based models [2, 4, 6, 10]. In this class of models, in addition to offers giving to the competitor, each agent can exchange extra pieces of information which are called “arguments”. Then, main pieces of information plus these arguments are sent to the competitor to assess. These transferred arguments by the agent could be useful for changing the competitor’s beliefs, goals, and even can give him encouragement to accept the given offer. This act is done in a way that decreases the process time. Although this class of models has some complexities like the others, it is practical in limited information situations [5]. Considering the concepts of concession rate and time limitations, and according to the fact that two sides of the negotiation are unaware of one another’s deadlines, in this paper we deal with an argument-based model of negotiation for multi-agent systems with incomplete information. The paper consists of following parts: In section 2 concepts of one-issue and multi-issues negotiation and optimal agenda are discussed. Section 3 is dedicated to introducing the proposed model for multi-issue negotiations, and eventually conclusions and future works are discussed in section 4. 2. ONE-ISSUE AND MULTI-ISSUES NEGOTIATION AND OPTIMAL AGENDA Multi-agent negotiations are based on one or multi issues [9]. Most researches in this field are dedicated to negotiations with only one issue (for example: price). However, in the real world most of negotiations and trades are multi-issues. In recent years, there have been few researches in this new field which have led to the formation of two fundamental concepts as follow [3,9]: a) Negotiation procedure: which clarifies the issues to be negotiated as a package simultaneously, or separately and in different times b) Agenda: which identifies the set of issues under negotiation (we call it common agenda sometimes). There is a common question among different researches on agenda which refers to defining the best candidates for each agent. In other words, it evaluates which candidate provides more utility for each agent and will be located in its optimal agenda? The answer is available in [9] and helps agents to select the proper issues in a negotiation. Based on this research, for each issue in the agenda the participating agents in a negotiation may have or have not a zone of agreement. and , in a situation where two • Note: In [9] , the optimal agendas for two negotiating agents and , respectively. Because of existence issues A and B are candidate to be negotiated, are shown by of two candidates issues for negotiation , four negotiation scenario are possible: o S1: both issues have a zone of agreement. o S2: only issue A has a zone of agreement. o S3: only issue B has a zone of agreement. o S4: neither issue A nor issue B has a zone of agreement. It is clear that among these four scenarios, in scenario S1 and S4, agents’ optimal agendas are the same, and they consist of either issues or none of them. But, for two other agents in scenarios S2 and S3 negotiating is useful only about one issue (A or B) for both sides, and optimal agendas are different. Optimal agendas of two agents for two different cases in scenario S2 are shown below (In figure 1 and 2 which are related to part a and b of this scenario , the utility boundaries for issues A and B are called and respectively and horizontal axis show agent ’s utility and vertical axis shows agents ‘s utility for these two issues). a) As it is clear in figure 1, issue B has positive utility for none of the agents (is placed in quadrant Q3). Apparently, none of the agents accepts negotiating about B in this case and optimal agendas of two agents just consist of issue {A}. b) Another case is that Issue B has positive utility for one of the agents (is placed in quadrant Q2 or Q4). It is predictable that in this case, only the agent that receives positive utility from negotiating on B , adds it to its optimal agenda. Therefore the optimal agendas will be in one of these two forms : 200 IADIS International Conference e-Commerce 2009 Figure 1. Agents Utilities in Scenario S2 (Part a) For example in figure 2 , only agent agendas are: Figure 2. Agents Utilities in Scenario S2 (Part b) gains a positive utility from negotiating on issue B and the optimal Definition: two agents optimal agendas in the above sample are {A,B} and {A} respectively, and the common available issue in both agents optimal agendas is {A}, and we call {B} the peripheral issue of the negotiation. According to the mentioned definition, two agents in scenarios such as S4 have no common issue, and all the issues in the agenda are peripheral issues. On the other hand, our proposed model of negotiation is mainly based on applying common issues (we will see it later); therefore, it is useable for all the cases except scenario S4. We should note that situations such as S4 are so rare–because most of the times at least one issue (such as price which is a common issue for both agents) exists in the e-commerce negotiations. As a result, this limitation of proposed model is not an important one, and traditional methods could be utilized if it was necessary. 3. THE PROPOSED MODEL Proposed model of this paper consists of five main steps which are discussed in this part and are traceable using figure 3. Step 1) At first existence of any issue in common agenda is assessed by the beginner. If no issue exists, this negotiation is related to scenario S4 and is out of the scope of this paper. As a result, the negotiation will be accomplished by using the old techniques if it is possible. Otherwise, if such issues exist , one side of the negotiation which we call it I ( beginner ) , will choose one of these common issues (maybe the one which can bring him more utility) and by giving his offer about the common issue to his competitor (J) , starts the negotiation (bold arrow in figure 4). Step 2) If J accepts the offer, negotiation will end successfully, otherwise J will compute its peripheral issues utilities and analyzes whether there are one or more peripheral issues available in his optimal agenda, that their entrance to the negotiation process, not only will compensate utility loss came through acceptance of I’s offer about the common issue, but also will be the best possible case (competitive) .Then, relation 1 should be validated. U total (t) ≥ u(t+1), u(t+2),… In relation 1, U total(t) Is the total utility of agent j for issues that are under negotiation at time t (common+ peripheral issues) and u(t+1), u(t+2),… are next cases utilities gained through negotiation about the common issue (our model is based on using common issues) at following times .The reason for this question is that agents will never accept utility loss. Therefore, they must be assured that it is not possible to gain more utility in future. On the other hand, due to concession rate principal, relation 2 is satisfied: (2) u(t+1) ≥u(t+2) , u(t+3),… 201 ISBN: 978-972-8924-89-8 © 2009 IADIS Considering relation 1 and 2, it is enough to compare the total utility with the utility gained through the negotiation about common issue in the next time of negotiation and relation 3 be validated. U total(t) ≥ u(t+1) )3( And based on definition of U total(t) relation 4 forms: (4) U total(t) = f pjj(t) +u(t) Relation 3 and 4 result in formation of relation 5 : f pjj(t) + u(t) ≥ u(t+1) (5) Figure 3. Flowchart of the Model fpjj(t) is sum of utilities, agent J ( first j in statement) Again, consider figure 4. In relations 4 and 5, gained through its own peripheral issues (second j in statement) at time t . u(t) is the utility that agent J gains through agent I’s offer about common issue ( I is proposer agent ). In all above relations, u(t+1) is the utility that J gains through its own offer about common issue at the next time of the negotiation; since, it is his turn to offer common issue at time t+1 . By considering uxy(t) as the utility, agent x gains through agent y’s offer about common issue at time t, relation 5 converts to relation 6: fpjj(t) + uji(t) ≥ ujj(t+1) Then, it is enough to evaluate correctness of relation 6 in step 2. Step 3) If peripheral issues that make relation 6 valid do not exist, by jumping to step 1 , agent J presents its counter offer to I (figure 3 simplifies the model trace by name exchanging of I and J). Otherwise if such issues exist, all possible sets of these issues will be sent from J to I at time t+1 (see figure 5). Then using an argument-based method, agent J announces I , that it will accept his offer about common issue (cooperative) If agent I agrees at least to one of the sent sets. Simultaneously, J will announce I , what would be his next counter offer about common issue, if I does not accept his offer about common issue. 202 IADIS International Conference e-Commerce 2009 Figure 4. Negotiation Process Up to the Second Step Figure 5. Negotiation Process in Steps 3 and 4 Step 4) At first competitor agent (I) assesses J’s counter offer about common issue. If accepts it, after Elimination of common issue from agenda, the negotiation will end successfully (This elimination is essential for preventing of repetitive negotiation for the same issue in future negotiations). Otherwise, J’s offers about peripheral issues are assessed and it is analyzed if the utility loss resulted from accepting J’s offers plus the utility that J accepts to concede I about common issue, is equal or less than best next case (the case that I’s next offer is accepted). Therefore relation 7 should be validated for agent I (figure 5) . fpij(t) + uii(t) ≥ uii(t+2) The reason for usage of uii(t) in the above formula is that ,when J at time t+1 announces accepting I’s offer under conditions, he means I’s offer at time t (because relation 6 was assessed in that time) and condition is that I accept one set of J’s offers about peripheral issues in that time too (fpij(t)) . Figure 5 shows what is happening in this step and how relation 7 is analyzed. In this figure, bold arrow is where we are now (time t+1) and shows the offer sent from J to I. Step 5) If relation 7 is not satisfied, again two agents’ roles will replace and algorithm jumps to step 1. But, if relation 1 is valid , ( after removing common issue and peripheral issues from agendas) the negotiation will end successfully. • Note: New negotiation processes may start by applying other common or peripheral issues. It seems that where in relation 7, n>0 ( some peripheral issues exist) and utility loss of the common issue between time t and t+2 , is greater than overall utility of peripheral issues at time t , the problem converges to the solution quickly ; The reason of this matter will be analyzed in the following part . At first , consider relation 7: fpij(t) ≥ uii(t+2)- uii(t) Then, based on the concession rate principal, right part of the above relation is negative and the left part is too; because utility of I’s peripheral issues are negative for J to a certainty. Therefore, relation 8 comes from relation 7: (8) | fpij(t)| ≤ |uii(t+2)- uii(t)| Figure 6. Final Task and Successful End of Negotiation Figure 6, demonstrates relation 8 operation and optimal negotiation process. In this figure , symbol | | shows passing some period of time which two agents didn’t reach agreement . Utility gained for agent I through negotiation about common issue at time t and t+2 are represented by black circles and Uii(t) and 203 ISBN: 978-972-8924-89-8 © 2009 IADIS Uii(t+2) symbols respectively. Utility gained through negotiation about peripheral issues at time t, is fpij(t) symbol. As it is illustrated In spite of negative utilities of J’s represented by white circle and peripheral issues for agent I (white circle), summation of these issues utilities (represented by Celtic cross) plus common issue utility at time t is greater than utility gained through negotiation about common issue at time t+2(Uii(t+2) ) . As a result, agent I accepts J’s offer about peripheral issues and negotiation ends successfully and in a shorter time ( As it is clear , in situation except the above case, the negotiation process continues) in addition to gaining higher utilities. 4. CONCLUSION As we know, by considering the concept of concession rate and time limitations for two negotiators and because of the fact that two sides of negotiation are usually unaware of each other deadlines, reaching quicker agreement is discussed as a fundamental problem and is the focus of lots of researches. Therefore, in this paper by applying peripheral issues and argument-based methods, a model of negotiation is presented which is used for reaching faster agreement and gaining more utility in some occasions. By applying such a competitive-cooperative method, this model uses the benefits of both competitive models (to increase utility) and cooperative ones (to reach quicker agreements). The proposed model can be improved in several main directions: At first, the limitations of the proposed model can be analyzed in more details. Secondly, the trust problem and the way of achieving it through the proposed model and how the agents should evaluate their competitor’s trustworthy about peripheral issues can be analyzed in more details. Thirdly, computational load of the proposed model can be assessed, specifically in situations where various and huge amounts of offers about peripheral issues are proposed from one agent side to the other, and the other has to figure out the utility of these issues in order to decide to refuse them totally or accept one at least. REFERENCES [1] sabyasachi S. et al , 2005 . a bayes net approach to argumentation based negotiation. springer publication argumentation in Multi-agent systems ، pp. 111-129. [2] C. Sierra .et al , 1998 . a framework for argumentation-based negotiation . springer publication , intelligent agent IV: 4th international workshop on Agent theories , architectures and languages (ATAL-1997) , pp.177-192. [3] S. Fatima .et al , 2004 . An agenda based framework for multi-issue negotiation . Artificial Intelligence Journal, 152(1):1–45 . [4] N.C. Karunatillake1 .et al , 2005 . Argument-Based Negotiation in a Social Context . Proceedings of the four international joint conference on Autonomous agents and multi-agent systems , pp.1331-1332. [5] N.R. Jennings .et al , 2001 . automated negotiation: prospect, methods and challenges . international journal of group decision and negotiation ,10(2):199.215. [6] L. Amgoud and C. Cayrol , 2002 . a reasoning model based on the production of acceptable arguments. annals of mathematics and artificial intelligence. [7] A. Pirvanescu .et al , 2005. DEVELOPING A JADE-BASED MULTI-AGENT E-COMMERCE ENVIRONMENT. Proceedings IADIS AC’05, International Conference on Applied Computing . Algarve, Portugal, pp. 425-432. [8] S. Buffett and B. Spencer , 2005 , Learning Opponents' Preferences in Multi-Object Automated Negotiation , In Proc. of the 7th International Conference on Electronic Commerce (ICEC'05), Xi’an , China , pp. 300-305. [9] S. Fatima .et al , 2004 . Optimal Negotiation of Multiple Issues in Incomplete Information Setting . Proceedings of the Third International Joint Conference on Autonomous Agents and Multi-agent Systems , vol. 3 , pp. 1080-1087. [10] Rahwan I. et al , 2003 . Towards Interest-Based Negotiation . International Conference on Autonomous Agents Proceedings of the second international joint conference on Autonomous agents and multi-agent systems , pp. 773780 [11] http://en.wikipedia.org/wiki/E-commerce 204 IADIS International Conference e-Commerce 2009 SELF-PRODUCT CONGRUENCE: IMAGE-PERCEPTIONS OF POSTMODERN OUTDOOR-APPAREL CONSUMERS IN E-COMMUNITIES Jan Breitsohl Bangor University Marwan Khammash Bangor University ABSTRACT The purpose of this paper is to revise the self-product congruence paradigm from a postmodern perspective. Self-product congruity (SPC) has been studied in various contexts, yet its theoretical foundation (i.e. the actual and ideal self-image) has not been challenged for the last decade, despite the widely-noticed transformation of global consumerism in consumer behavior literature. Looking at outdoor-apparel consumption in e-communities, this paper aims to explore the postmodern role of actual- and ideal self-product congruence theory. It is hypothesized that the actual SPC of those engaged in outdoor activities (here named ‘enthusiasts’) is more closely related to performance attributes whereas their ideal SPC relates stronger to fashion-appeals. For those fashion-conscious consumers not active in the outdoors (here called ‘Fashionists’), the vice versa is expected. Findings may contribute to the current academic debate on viewing selfproduct-congruence as a holistic (i.e. stereotyping) rather than multidimensional concept. Expected relevant results for online/offline marketing practitioners are the necessary transformation of the current fashion-performance dichotomy in outdoor apparel segmentation and positioning strategies. Additional managerial implications of postmodern consumption behavior in e-communities are discussed. KEYWORDS Self-Product Congruence, Postmodern Consumption, Consumer Behavior, e-Communities, Outdoor-Apparel. 1. INTRODUCTION Since it was first brought to the attention of consumer behavior analysts almost half a century ago (e.g. Birdwell 1964), self-product congruence (SPC) has experienced an ongoing academic discourse. At its core, SPC describes the match between a product’s image and a consumer’s self-image. Product image is often used synonymously with brand personality or brand-image and describes a set of personality attributes associated with a product (Aaker 1997). Self-image may simply be defined as the perception’s one has about himself (Sirgy 1982), commonly equated with the term self-image. Whereas early studies focused on a one-dimensional SPC paradigm, later research (e.g. Ross 1971) applied the established distinction between one’s actual- (the way we see ourselves) and ideal self-image (the way we would like to be). Despite a continuous expansion of the paradigm– proposing malleable, multiple or social self-images – the original two-dimensional SPC framework has not been revised alongside the increasing depth of postmodern marketing research (see Firat & Schultz 2001). Contemporary, postmodernist school of thought proposes a fragmentation of one’s self within a pluralist society where self-images are dynamic, individualized and under constant pressure to be expressed and communicated via ones consumption patterns. Thus, the multidimensional conflict between one’s actual-and ideal self (Belk 1988) can be expected to have changed the stereotypical perception of a product-user image as basis of comparison for self-product congruence. In the context of the analyzed outdoor-apparel industry, this phenomenon is expected to occur for the two industry segments currently served by outdoor-apparel companies: Outdoor- Enthusiasts who engage in nature-activities and Fashion-consumers (Fashionists) who mainly wear the apparel in an urban environment. Enthusiasts’ actual SPC can be expected to be based on physical activity and adventures in the outdoors, 205 ISBN: 978-972-8924-89-8 © 2009 IADIS whereas Fashionists may rather focus on hedonistic and aesthetic appeals. However, the postmodern environment may see Enthusiasts aspiring to be fashionable and hence to be integrated in urban aesthetics, whereas a Fashionist’s ideal may revolve around escaping consumption pressures and seeking for authentic nature experiences. Consequently, online marketers will be well-advised to revise their contemporary segmentation profile- and positioning strategies. 2. PROBLEM STATEMENT A dominant number of SPC scholars equate product-image with product user image, hence linking the concept to ‘typical users’ or ‘prototypes’ (see for example Sirgy 1982; Sirgy et al. 1997). With regards to the aforementioned potential transformation of the two-dimensional self-image paradigm, this stereotyping may be a sign of marketing myopia, i.e. being short-sighted in terms of customer-orientation. The current targeting strategies cultivate outdoor-apparel branding in a dominantly adventurous, extreme-sports context or, more recently, with a fashion conscious, urban focus. Yet, the validity of such prototyping has not been verified with regards to postmodern SPC considerations and may therefore necessitate a revision of the current theoretical construct. 3. HYPOTHESIS The central objective of this study is to analyze the validity of using stereotypical product user-images for two dimensions (actual and ideal) of the SPC-model. The outdoor apparel industry currently proposes two stereotypes: Performance-oriented nature enthusiasts, who actually engage in outdoor activities (such as hiking, climbing etc.) and fashion-concerned urban consumers who are not interested in product performance as they do not actively pursue those activities. To analyze the strength of this typifying assumption, the following hypothesis is proposed: H1: The relation between the Enthusiasts’ actual SPC and product-image attributes is stronger than the relation between the Fashionists’ SPC and product-image attributes Yet, these suppositions may only offer a fragmented picture of the multidimensional self-images these two industry segments experience. Contemporary consumptions cultures are pluralistic and demand constant symbolic interactionism in different contexts. Henceforth, an aspiration for ‘symbolic integration’ as much as a perceived need for ‘authentic utilitarianism’ may result in ideal self-images that significantly divert from actual self-images. Marketers’ linear assumptions of extreme adventures and fashion trends as ideal images may require a revision that is encapsulated in the following hypotheses: H2: The relation between the Enthusiast’s ideal SPC and product-image attributes is weaker than the relation between the Fashionist’s ideal SPC and product-image attributes. 4. RESEARCH IMPORTANCE The dilemma postmodern consumerism represents for marketers is to develop dynamic segmentation strategies that coherently sketch increasingly heterogeneous target groups. The fragmentation of the self and hyper-realities (e.g. conflicting role expectations or virtual worlds) in e-communities have made stereotyping consumer clusters with clearly definable product images a highly complex, if not misguided task. Investigating SPC from a postmodern perspective is expected to reveal that consumers are more selfaware (Hogg et al 2000) and therefore clearly distinguish between the two SPC dimensions (actual and ideal).This will further highlight the relevance of marketing approaches such as self-esteem advertising, positioning strategies and celebrity endorsement precision Equally important, scholars have found that SPC positively impacts on marketing variables such as brand loyalty (Kressmann 2006), purchase intentions (Sirgy 1982) and brand preference (Aaker 1999).The psychoanalytical nature of the SPC paradigm can be meaningfully related to any marketing factors and hence a 206 IADIS International Conference e-Commerce 2009 revision of its underlying principles – as is done in this paper with regards to actual and ideal SPC- paves the way for valid future research projects. In case of results contrary to the expected, this study may still indicate the importance of further research: If a stereotypical segmentation for both actual and ideal SPC is supported, the investigation of whether multiple selves equally hold for this conclusion will be of insightful use for scholars and marketers. Outdoor apparel marketers will acquire a theoretical verification for their current targeting strategies which have received little academic attention so far. 5. RESEARCH METHODS 5.1 Product Stimulus The chosen product category is outdoor apparel, which will here be defined as functional clothing and footwear for the nature-activities hiking, climbing, mountaineering, mountain biking and cross-country running. 5.2 Sampling This paper will use an online-survey research strategy. A brief introductory note regarding purpose, length and significance of the survey together with a web-link to the survey will be placed in 2 e-communities: Outdoorsmagic.com: (an Online forum focusing on outdoor activities - 20,000 members) and Xing.com: Group ‘Outdoor’ (8,800 members interested in outdoor sports / activities) Around 300 replies are anticipated of which 200 are hoped to be complete and usable. In case the respondent rate is lower than expected, further online surveys will be conducted with a convenience sample of university students. Online surveys are commonly used in SCP research and have generally produced sufficient sample sizes in comparable studies (e.g. Krohmer 2007). No further demographic segmentation will be made. 5.3 Data Collection Data will be gathered using questionnaires created with the survey-tool ‘Surveymonkey’ (see surveymonkey.com). Participants will be asked to complete the questionnaire with regards to their own outdoor apparel. The possession of outdoor apparel will also function as a required pre-selection condition for the inclusion of a respondent in the data analysis. 5.4 Questionnaire Design This paper will adopt Kressmann et al.’s (2006) SPC-questionnaire. The structure will, after a brief introduction (purpose, structure, duration), run as follows: Section 1 – Screening basic information regarding participants’ outdoor apparel consumption. Section 2 – Personality attributes ‘How important is it for you that a person is.....?’ Participants are required to indicate the personal importance put on 15 given personality facets (e.g. tough, up-to-date, honest). Section 3 – Actual self-image ‘To what extent do the following personality attributes apply to you? :’ Section 4 – Ideal self-image ‘Imagine how you would like to be. To what extent do the following personality attributes apply to how you would like to be?’ Section 5 – Product image ‘Image your brand of outdoor apparel (e.g. The North Face) as a person. Indicate the extent to which the following personality attributes apply to the typical user of the brand.’ This question originally read ’apply to the brand or the typical user of the brand’. Due to the aforementioned doubts regarding the validity of equating product-image with the image of typical productusers, ‘...the brand or...’ was excluded from the question. Answers may therefore only represent a common, stereotyped product-user image and not individualistic perceptions of a desired (ideal) product image. 207 ISBN: 978-972-8924-89-8 © 2009 IADIS The 15 items will be based on Aaker’s (1997) brand personality measure and used continuously for section 2-5. Some items will be modified to reflect the postmodern outdoor apparel paradigm of this paper. Answers will be measured using a 7-point-Likert-scale (from ‘completely unimportant’ (1) to ‘very important’ (7). Section 6 – Demographic information (asked at the end to avoid fatigue bias). The questionnaire design approaches SPC by investigating self-image and product image separately for a subsequent mathematical combination, a method long established in the SPC literature. 5.5 Data Analysis Data will be analyzed using SPSS statistical analysis software. Self-product congruity scores (Dk) will be measured using absolute differences scores between each brand personality rating and its correspondent selfimage ratings (actual and ideal) and will then be averaged across all personality attributes for each respondent (Sirgy, 1982; Sirgy et al., 1991,1997): n number of personality attributes (n=15), i personality attribute i (i=1...n), BPik brand rating along personality attribute i for respondent k, SIik self-image (actual or ideal) rating along personality attribute i for respondent k. To separately investigate both actual and ideal self-congruity, this formula will be further operationalised into the two indices. In addition, the personal importance of each brand personality item will be integrated into each of the indices and the index multiplied with -1 so that the larger values would indicate high selfcongruity and vice versa (Kressmann 2006): Wik importance rating of personality attribute i for respondent k; ASCik actual self-congruity (average weighted congruity between brand personality and actual self-image) for respondent k; ISCik ideal selfcongruity (average weighted congruity between brand personality and ideal self-image) for respondent k; ASIik actual self-image rating of personality attribute i for respondent k; ISCik ideal self-image rating of attribute i for respondent k. 6. CONCLUSION & LIMITATIONS Findings may contribute to the current academic debate on viewing self-product-congruence as a holistic (i.e. stereotyping) rather than multidimensional concept. Expected relevant results for online/offline marketing practitioners are the necessary transformation of the current fashion-performance dichotomy in outdoor apparel segmentation and positioning strategies. However, cultural differences which may lead to different SPC relationships (Quester 2000) are not taken into account. In addition, multiple selves beyond the used two-dimensional SPC construct are also not taken into consideration. Finally, functional congruity, which is shown to be dominated by SPC, (Sirgy 1991) is not included but may be a relevant area for further research with regards to the performance attributes of outdoor apparel. 208 IADIS International Conference e-Commerce 2009 REFERENCES Aaker, JL 1997, ‘Dimensions of brand personality’, Journal of Marketing Research, vol. 34, no. 3, pp. 347-356. Aaker, JL 1999, ‘The malleable self: the role of self-expression in persuasion’, Journal of Marketing Research, vol. 36, pp. 45-57. Belk, RW 1988, ‘Possessions and the extended self’, Journal of Consumer Research, vol. 15, no. 2, pp. 139-168. Birdwell, AL E 1968, ‘A study of the influence of image congruence on consumer choice’, The Journal of Business, vol. 41, no. 1, pp. 17-88. Firat, Fuat A & Clifford J Shultz 2001, “Preliminary Metric. Investigations into the Postmodern Consumer,” Marketing. Letters, vol. 12, no. 2, pp. 189- 203. Hogg, MK, Cox, AJ, & Kneeling, K 2000, ‘The impact of self-monitoring on image congruence and product/brand evaluation’, European Journal of Marketing, vol. 34, no. 5/6, pp. 641-666. Kressmann, F, Sirgy, MJ, Herrmann, A, Huber, F, Huber, S, & Lee, DJ 2006, ‘Direct and indirect effects of self-image congruence on brand loyalty’, Journal of Business Research, vol. 59, pp. 955-964. Krohmer, H, Malaer, L & Nyffenegger, B 2007, ‘The fit between brand personality and consumer self: the importance of actual versus ideal self for brand performance’, Proceedings of the AMA Winter Educators’ Conference Brand Identity Communications, San Diego, USA, pp. 1-20. Quester, PG, Karunaratna, A, & Goh LK 2000, ’Self-congruity and product evaluation: across-cultural study’, Journal of Consumer Marketing, vol. 17, no. 6, pp. 525-537. Ross, I 1971, Self-concept and brand preference’, Journal of business, vol. 44, no. 1, pp. 38- 50. Sirgy, MJ 1982, ‘Self-concept in consumer behaviour: a critical review’, Journal of Consumer Research, vol. 9, no. 3, pp. 287-300. Sirgy, MJ, Johar, JS, Samli, AC, & Caliborine 1991, ‘Self-congruity versus functional congruity: predictors of consumer behavior’, Journal of the Academy of Marketing Science, vol. 19, no. 4, pp. 363-375. Sirgy, MJ, Grewal, D, Mangleburg, Park, JO, Chon, KS, Claiborne, CB, Johar, JS, & Berkman 1997, ‘Assessing the predictive validity of two methods of measuring self-image congruence’, Journal of the Academy of Marketing Science, vol. 25, no. 3, pp. 229-241. 209 ISBN: 978-972-8924-89-8 © 2009 IADIS EXTERNALIZATION OF VIRTUAL PROTOTYPES AS AN E-COMMERCE SERVICE IN THE FASHION INDUSTRY Carolin Löffler Institute for Information Systems II, University of Erlangen-Nuremberg, Lange Gasse 20, 90403 Nuremberg, Germany ABSTRACT Due to a lack of expertise on how to design appropriate support of product lifecycle processes more efficiently, companies increasingly have difficulties identifying and using potentials for optimization. This research project aims at the development of a new concept that generates added services for customers, suppliers and partners and – at the same time – supports the product lifecycle management of a company. This approach is illustrated by an action research project in the fashion industry. KEYWORDS E-Commerce, virtual prototyping, service management, product lifecycle management, added services. 1. INTRODUCTION In today’s global business the big challenge for international companies is to generate a competitive advantage and growth potential. Many markets are nearly saturated and for reasons of economy products are standardized and similar in quality and technology between the competitors [Welge and Holtbrügge 2006]. Due to the intensive competition on international markets, it is getting more and more difficult for companies to distinguish themselves with their core products and to safeguard market share in the long run [Piller 2000]. Product-supporting services – so called „added services“ – help companies to differentiate their product range from the competitor´s and to reach a secure market position by offering integral product packages [Cassack 2005]. Despite this knowledge there is no strategic development of product-supported services. Most added services are developed only on customer inquiry [Storey and Kelly 2001]. The customer-oriented approach deals with the client’s expectations and individual conveniences. First the customer’s special requirements are identified and then the single elements of service on a customer-specific basis are developed. Bruhn (2001) states, that this is the best way to gain business success. However, many added service rollouts fail despite the knowledge that they would generate a competitive advantage and customer loyalty [Hemetsberger and Füller 2006]. The crucial point is that for the decision maker in a company, the organizational efficiency is more important than customer-oriented development of services. When it turns out, that a new customer-oriented added service is too costly, there is less to gain for the company [Holtgrewe and Kerst 2002]. This research project “service externalization” focuses on the development of new added services for ecommerce based on IT-driven services within the product lifecycle management. 2. CURRENT STATE OF RESEARCH The term “service externalization” is used by Meyer and Noch in 1992: „With externalization you can create an external service out of an internal one“ [Meyer and Noch 1992, Noch 1995]. In later publications Meyer is talking about externalization only as transfer of manpower from the service producer to the consumer [Meyer 2004]. Lorenz-Meyer (2004) picked up on the primary idea of service externalization in terms of Meyer and 210 IADIS International Conference e-Commerce 2009 Noch and distanced himself from Corsten´s approach1. But he does not address to how an externalization works and which aspects should be considered. He simply gives the hint, that services applied in human resources and finance should be examined focusing costs and competences and to what extent the services are appropriate for an external offering [Lorenz-Meyer 2004]. The topic service externalization as a new approach to develop services is only analyzed shortly in the literature. This research project helps to close the identified research gap by analyzing the approach more precisely with the help of an action research project in the fashion industry. 3. SERVICE EXTERNALIZATION Service Externalization (ServEx) is a concept within service management. It marks a systematic approach to identify high potential internal services within the product lifecycle of a company and to offer those services also to other departments, to customers and suppliers of a company. Compared to the traditional customer-oriented view the concept of service externalization describes a completely different approach: Instead of an outside-in process (starting from the customer) we talk about an inside-out process, which can be distinguished in an internal (within the company) and an external (to the partners of the company) externalization process (Figure 1): Figure 1. Service Externalization Internal externalization process: In service externalization within the company, new services based on an internal one are developed. There is the possibility to modify the original service into different specifications. It can be either partly or entirely changed, modified into a new service and be either appointed for the same or a different function. Apart for the previous internal users, the service can be used in other parts of the company as well. The application of the new service is therefore only limited to the company. External externalization process: Analog to service externalization within the company - service externalization can also be applied to the partners of the company. The result of this procedure is a new service for customers, suppliers or partners. 1 Corsten´s approach of service externalization means the integration of an external factor (the customer) into the service compilation process. This procedure aims at saving costs with customer integration. Customer integration in these terms means transferring parts of the goods and services process to the consumer [Corsten 2000]. 211 ISBN: 978-972-8924-89-8 © 2009 IADIS 4. VIRTUAL PROTOTYPES AS A SERVICE IN THE FASHION INDUSTRY Virtual prototypes as an internal service are mainly used for product development in the fields of industrial engineering. Experts hope to reduce the product development time [Niesing and Miller 2001]. In a previous project we introduced virtual prototypes as a service also to the product development process of a fashion manufacturing company. In the following action research project we first take a look at virtual prototypes in the product development processes in the fashion industry and figure out, if virtual prototype services can be applied to other departments of the fashion manufacturing company. In the second step we take a look at the externalization potential of virtual prototype services towards the partners of the company. 4.1 Research Method The study reported here was designed to identify whether virtual prototyping can be integrated successfully in the product development process in the fashion industry. Furthermore it shows how to integrate virtual prototypes into other departments of the company and finally how to externalize parts of the solution as added services to customer, supplier or partner. The chosen research methodology is action research. A team of researchers worked for one year together with the IT innovation team of the fashion manufacturing company. The chosen company operates in the sporting goods manufacturing industry, has about 30.000 employees, and a 10 billion € turnover. For the study 15 managers and experts were interviewed and we took part in 8 workshops about 3D technologies. 4.2 Virtual Prototypes A virtual prototype is a computer based illustration of a future product [Cassack 2006]. The conventional designs on paper are the basis for likewise two dimensional computer based CAD drawings or technical designs. On this foundation the perspective is expanded to the third dimension that makes a 3D demonstration possible and let object seem real. 4.3 Application in the Product Development Process On this basis LaBat and Sokolowski (1999) summed up the product development process as following: At the start of the product development is the “problem definition and research” phase. After the target group for the range has been decided upon, the new properties and requirements for functions and design are defined and determined. Thereafter follows the “creative exploration” phase. This includes the idea generation and the conceptual design, design changes, prototype development and the rating of ideas and prototypes. From the existing product ideas, those are selected that eventually should hit the market in the “implementation phase”. Among others the production costs and the sequence of production is determined as well as a last fine tuning of the product is carried out. Through the application of virtual models, the development process can be speeded up, because physical samples are omitted. For the creation of virtual prototypes new internal costs emerge, which can be offset through the rationalization potential of 3D models. In the first design examination meeting, virtual models allow for a better visualization of the designed products. Coloring, material properties and cut can be illustrated more realistically than a print out or sketch by the designer. That way design aberrations that normally are only discovered with physical samples in later phases can be discovered. A further advantage is that changes can be directly made to the 3D model. Therefore new physical samples for the verification are no longer needed and sample rounds are reduced. The waiting time of up to 3 weeks that result through the production and shipment of the samples is omitted. Within a short time the designer can create many variations directly on the 3D model and receive a new visual impression in real time. 212 IADIS International Conference e-Commerce 2009 4.4 Service Externalization Company Services Sales representatives usually sell their products to retail using a product catalog and the help of physical samples. This process takes places around six to eight weeks prior to the official launch of a new range. Large sports goods producers such as adidas AG or Nike INC have two main catalogs per year in which several thousand products are presented. Large fashion companies partly have more than 30.000 pieces of clothes in their selection. When creating pictures for the pieces of clothes there are generally four different approaches: Either the collection is photographed using with models or two dimensional sketches are made. The photography allows for a realistic illustration of the range (e.g. fitting) and ensures a consistent demonstration format in the catalog. For this method a physical sample is needed for every piece of clothing. Because of the time intensive production of the samples and the photography of the range a delay in the catalog time table is possible. The catalog sketches that are created on the basis of the product description and the preliminary designs can already be produced without a physical sample. These pictures only communicate an incomplete impression of what the product will look like later. Material properties and the surface can only be illustrated in a limited manner. On the base of 3D models it is possible to generate virtual product photos. Therefore the catalog’s picture requirements (angle, shadows, file format) need to be precisely specified. These settings are managed in the software and with the press of a button or fully automatically catalog pictures can be created. This procedure combines the advantages of sketch and photography. High quality catalog pictures can be created without the time intensive use of physical samples. Customer Services The sale of fashion items is almost exclusively transacted through business customers. These business customers can for example be sports item dealers (e. g. Footlocker) or large department stores (Marcy´s, Wal-Mart). Usually these create the range illustrations for their own catalogs. This is normally because suppliers prefer an individual illustration style and use these for advertising reasons, e.g. in catalogs. Therefore the arrangement of the products, perspectives and background can vary. The in-house catalog created is cost and time intensive and product can party only be advertised and sold with a delay. Company can offer its B2B customers individualized product pictures for their catalogues, therefore only a specification of the customer’s requests concerning the illustration style of the products in the catalog and its implementation into the system is necessary. If the customer is furthermore offered an interface, the settings of every product can be individually changed. The creation of private avatars with pre defined measures is also conceivable. The business customer can from then on separately configure the type of illustration (e.g. with or without avatar) for every item. Supplier Services Virtual prototypes can be used for material and supply planning. The suppliers can in an early product development phase be informed in detail about the later product, so these can ideally order, select and adjust their materials. When creating the samples it is often the case that these don’t match the planned products. This can have different reasons: The supplier for example adopts a wrong fit or the product description isn’t precise enough. If a virtual model is available at an early stage these mistakes can be reduced because the manufacturer gets a clear picture of the future product. Partner Services Normally the external partner’s salesman takes all the available items of a product in all available colors with him during a customer call (e.g. sports item dealer, department stores). If instead the salesman is additionally equipped with virtual prototypes, the so called virtual sales man samples, the number of physical samples is considerably reduced. It is sufficient to take on only a single physical exemplar. With the help of a note book and a high-grade projector all models in their available color combinations can be presented to the customer. A physical sample is though still need, because the haptic cognition still plays an important role in the sales process. 213 ISBN: 978-972-8924-89-8 © 2009 IADIS 5. CONCLUSION AND FURTHER RESEARCH This research project has illustrated a procedure that shows that virtual prototype services in the fashion industry can be successfully applied in the product development process. Through externalization virtual prototyping can create additional value adds to the original intended use. In the further research, acceptance analyses were performed and a change management concept developed that prepares all parties concerned for the introduction of virtual prototypes. Likewise the transferability to other companies is analyzed and an ‘externalization road map’ of virtual prototype services developed. REFERENCES Bruhn, M., 2001. Relationship Marketing. Das Management der Kundenbeziehungen. München, Vahlen. Cassack, I., 2006. Prototypengestützte Kosten- und Erlösplanung für produktbegleitende Dienstleistungen. Wiesbaden, Gabler. Corsten, H., 2000. Der Integrationsgrad des externen Faktors als Gestaltungsparamenter in Dienstleistungsunternehmen. Voraussetzungen und Möglichkeiten der Externalisierung und Internalisierung. In M. Bruhn and B. Stauss, Dienstleistungsqualität. Konzepte - Methoden – Erfahrungen. Wiesbaden, Gabler, pp. 145-168. Hemetsberger, A. and Füller, J., 2006. Qual der Wahl - Welche Methode führt zu kundenorientierten Innovationen? In H. H. Hinterhuber and K. Matzler, Kundenorientierte Unternehmensführung, Kundenorientierung Kundenzufriedenheit – Kundenbindung. Wiesbaden, Gabler, pp. 399-433. Hotgrewe, U. and Kerst, C., 2002. Zwischen Kundenorientierung und organisatorischer Effizienz - Callcenter als Grenzstellen. Soziale Welt, Vol 53/2, pp. 141-160. LaBat, K. L. and Sokolowski, S. L., 1998. A three-stage design process applied to an industry university textile product design project. Clothing and Textiles Research Journal. Vol. 17/1, pp. 11-20. Lorenz-Meyer, D., 2004. Management industrieller Dienstleistungen. Ein Leitfaden zur effizienten Gestaltung von industriellen Dienstleistungsangeboten. Wiesbaden, Gabler. Meyer, A., 2004. Dienstleistungsmarketing. Impulse für Forschung und Management. Wiesbaden, Gabler. Meyer, A. and Noch, R., 1992. Dienstleistungen im Investitionsgütermarketing. Das Wirtschaftsstudium, Vol 12, pp. 954-961. Niesing, B. and Franz, M., 2001. Schneller, schneller, Virtual Prototyping. Retrieved 07 29, 2008, from Fraunhofer Institut: http://www.fraunhofer.de/archiv/magazin/pflege.zv.fhg.de/german/publications/df/df2001/mag3-2001 _42.pdf Noch, R., 1995. Dienstleistungen im Investitionsgüter-Marketing. Strategien und Umsetzung. München, FGM-Verlag. Piller, F., 2000. Mass Customization. Ein wettbewerbsstrategisches Konzept im Informationszeitalter. Wiesbaden, Gabler. Storey, C. and Kelly, D., 2001. Measuring Performance of New Service Development Activities. The Service Industries Journal, Vol 21/2, pp. 71-90. Welge, M. K. and Holtbrügge, D., 2006. Internationales Management - Theorien, Funktionen, Fallstudien. Stuttgart, Poeschel. 214 IADIS International Conference e-Commerce 2009 BUILDING FINANCIAL CAPABILITY VIA THE INTERNET Tanai Khiaonarong Westminster Business School, University of Westminster 35 Marylebone Road, London NW1 5LS, United Kingdom ABSTRACT This paper introduces a nodality framework to help assess financial capability building websites. First, it outlines the major driving forces behind financial capability building and reviews the use of electronic delivery channels. An analytical framework based on nodality as a tool in delivering e-government services is presented. We then present an illustrative case of the Financial Services Authority which leads the national strategy for financial capability building in the United Kingdom. We conclude by identifying some research issues. KEYWORDS Financial capability, Internet, e-government, nodality. 1. INTRODUCTION Financial capability development (also known as financial literacy and financial education) has been a major challenge faced by many countries around the world. Financial services have brought with them both opportunities and risks for a wide range of stakeholders including, international organisations, governments, central banks, financial regulators, financial institutions, non-profit organisations, consumers, and taxpayers. This challenge has become more immediate after the US sub-prime mortgage crisis where the lack of understanding in opaque financial instruments and their inherent risks led to a global credit crisis. The need to improve financial capability has therefore been of paramount importance and this has been driven by three major forces (OECD, 2005a, 2005b; Braunstein and Welch, 2002). First, financial products and services have become more complex and varied in number, which could pose risks to consumers, financial institutions, and above all, financial stability if not well understood or managed. While financial and technological innovation has accelerated with financial liberalization and deregulation, there remains a regulatory catch-up challenge in many countries where adequate risk management and consumer understanding of financial issues could be compromised. Second, aging societies, through the retirement of the baby boom generation and increases in life expectancy, would put pressure on public retirement programs. In this connection, the trend of changing pension arrangements from defined benefits to defined contributions has shifted risks from the provider to the worker as retirement income is no longer guaranteed but based on contribution rates and investment decisions made by a worker during his working life. And third, financial understanding among consumers was found to be low in a review of financial literacy surveys in twelve OECD countries. The range of educational facilities and services to support financial capability building can be varied as found in a recent study of central bank economic and financial literacy programs (Gnan, et al., 2007; Fluch, 2007). This includes money museums, open house programs, visitor centers, guided tours, training facilities, special seminars, competitions, fund raising events, and e-education resources. In this paper, we focus on the use of e-education resources, particularly the use of the Internet. For example, this has included the following: a global clearinghouse on financial education information, data, resources, research, programs and news around the world (OECD, 1999); e-government services (Australian Government, 2009; FSA, 2009; U.S. Financial Literacy and Education Commission, 2009); on-line study sites on monetary and financial matters with an interactive monetary policy game (Bank of Finland, 2009); and financial education material posted on video sharing websites like www.youtube.com (European Central Bank, 2009). 215 ISBN: 978-972-8924-89-8 © 2009 IADIS This paper aims to develop a framework for the analysis and assessment of financial capability building websites in the following section. We provide an illustrative case of the United Kingdom where one of the world’s largest international financial center is located and where a National Strategy for Financial Capability is led by the Financial Services Authority (FSA, 2004; 2006a; 2006b). We conclude with a discussion of further research issues, particularly on the use of structural and user metric analyses. 2. NODALITY FRAMEWORK We adopt the concept of nodality – the property of being at the centre of social and informational networks as a tool for assessing financial capability building on the Internet. Nodality has been a concept used in the study of governments and has more recently been applied to the analysis of e-government (Hood, 1983; Hood and Margetts, 2006; Escher et al., 2006; National Audit Office, 2007). In their study of the nodality of digital government focused on three foreign office websites (Australia, UK and US), Escher et al. (2006) note five dimensions where nodality can be measured as follows: 1) Visibility (likelihood of an organization being found from a search engine); 2) Accessibility (ease of obtaining information); 3) Navigability (ease of moving around site and finding related content); 4) Extroversion (number of other sites and sources of information to which users are directed once they are there); and 5) Competitiveness (competing against other sites which provide similar information). They found high accessibility and competitiveness for the Australian website, and also high visibility, extroversion and navigability for the UK website. To quantify the above dimensions, we develop some structural metrics (measure properties such as the average distance between two random pages and the interconnectedness of sites) to help assess the property of financial capability building websites, while a possible data collection method would be through web crawling (eg Nutch program). Table 1 outlines the nodality framework. Table 1. Nodality Framework Name Inlink analysis Inlink analysis Description How many links are there into the site? Which types of domain are linking to the sites – governmental, commercial or voluntary and from which countries? Are the links generic (to home page) or specific (to particular topic on site) – ‘deep’ links Directed average distance Average length of all shortest path Outlink analysis How many links to external sites are there? Which type of domains (governmental, commercial or voluntary) do sites link? Comparative analysis How does the website compare with other similar websites? Interpretation High numbers of in-links will raise site’s visibility on search engine listings – making the site more visible to users of external search engines Nodality Indicator Visibility ‘Deep links’ returned by search engines tend to be more useful – as take users directly to information required rather than depositing them at home page where they must navigate to find what they want – but note deep links more likely to break Indicator of how easy it is to navigate the site. Smaller values indicate it is easier to reach other relevant pages High numbers of external links indicate an ‘extroverted’ rather than ‘introverted’ site which leads citizens to a wide range of further information, enriching the interaction Possible indicator of reliability of information and reputability of service provider Accessibility Source: Adapted from Escher et al. (2006, pp. 10-11) 216 Navigability Extroversion Competitiveness IADIS International Conference e-Commerce 2009 3. CASE STUDY: FINANCIAL SERVICES AUTHORITY The Financial Services Authority (FSA) is the United Kingdom’s financial regulator and acts as an independent non-governmental body, given statutory powers by the Financial Services and Markets Act 2000. It forms part of the broader tripartite framework with Her Majesty’s (HM) Treasury and the Bank of England in the oversight of financial stability in the UK (Bank of England, 2009). The FSA leads the national strategy for financial capability building and works to improve the public’s understanding of personal finance. As of December 2008, the FSA has reached 5.6 million people out of its target of 10 million. A major part of this strategy is the use of the Internet to help disseminate electronic materials on financial education. This study builds on earlier exploratory, baseline, financial education evaluation, and behavioral economic studies (FSA, 2005; 2006c; 2008a; 2008b). In particular, it seeks to evaluate the effectiveness of on-line tools in financial capability building programs in line with follow-up action plans by the UK government (HM Treasury and FSA, 2008a), financial capability intermediate targets (2008-9) and overall targets (2006-11) (FSA, 2008c). We aim to apply the nodality framework to assess the on-line presence of the FSA website (www.fsa.gov.uk). Through structural metric analysis using a web crawling tool (eg. Nutch), we seek to analyze the website in terms of its visibility, accessibility, navigability and extroversion. Moreover, we aim to assess the on-line presence of similar financial capability building websites elsewhere to explain the fifth dimension – competitiveness. Table 2 outlines the country, departments and corresponding websites where there is a financial capability building program. Although each country program is unique, it helps illustrate progress towards financial capability building within the OECD group of countries. Table 2. Financial Capability Building Websites Country Australia Canada United States Department Financial Literacy Foundation, Department of Treasury Consumer Education Program and the Financial Literacy Initiative, The Financial Consumer Agency of Canada Financial Literacy Education Commission, US Treasury Office of Financial Education Domains http://www.understandingmoney.gov.au/ http://www.themoneybelt.gc.ca/ http://www.mymoney.gov/ 4. CONCLUSION AND FURTHER RESEARCH Going forward, we seek to crawl the FSA website using the Nutch program to analyse it structural metrics. We also plan to conduct a similar analysis of similar government websites in Australia, Canada and the United States. This would help assess practices in countries that have well established financial capability building programs, which may be further support the international principles and practices for promoting financial education and awareness (OECD, 2005a, 2005b). Lessons learned from this research may also contribute towards work on financial inclusion and international development. Although this research may be limited to government websites, financial capability building websites extend beyond that, and may include websites provided by the private sector and civil society. Thus, it would also be interesting to analyse how the nodality framework, which has the property of being at the centre of social and informational networks, could be applied in a non-governmental context. Nevertheless, the study does have limitations as it focuses on outreach via the Internet. Other deliver channels, particularly partnership programs with various stakeholders such as schools, young adults, new parents, higher educations and workplaces, do exist and can be very effective. This is noted in the study where applicable to provide an overall view of financial capability building. 217 ISBN: 978-972-8924-89-8 © 2009 IADIS REFERENCES Australian Government (2009) Welcome to the Understanding Money Website. Available from: http://www.understandingmoney.gov.au/ [Accessed date: 22 January 2009]. Bank of England (2009) Memorandum of Understanding between HM Treasury, the Bank of England and the Financial Services Authority. Available from: http://www.bankofengland.co.uk/financialstability/mou.pdf [Accessed date: 22 January 2009]. Bank of Finland (2009) Bank of Finland On-line Learning Package. Available from: http://www.euro.fi/e/index.html [Accessed date: 22 January 2009]. Braunstein, S., and Welch, C. (2002) Financial Literacy: An Overview of Practice, Research, and Policy. Federal Reserve Bulletin, November, 445-457. Escher, T., Margetts, H., Petricek, V., and Cox, I. (2006) Governing from the Centre? Comparing the Nodality of Digital Governments, Prepared for delivery at the 2006 Annual Meeting of the American Political Science Association, August 31 - September 4, 2006. European Central Bank (2009) European Central Bank Educational Video. Available from: http://uk.youtube.com/watch?v=7sjQ7ly2NDU [Accessed date: 22 January 2009]. Financial Services Authority (FSA) (2004) Building Financial Capability in the UK: the Role of Advice, July. _____ (2005) Measuring Financial Capability: An Exploratory Study, Consumer Research 37, Prepared for the Financial Services Authority by Personal Finance Research Centre, University of Bristol, June. _____ (2006a) Financial Capability in the UK: Delivering Change, March. _____ (2006b) Financial Capability in the UK: Establishing a Baseline, March. _____ (2006c) Levels of Financial Capability in the UK: Results of a Baseline Survey, Consumer Research 47, Prepared for the Financial Services Authority by Personal Finance Research Centre, University of Bristol, March. _____ (2008a) Evidence of Impact: An Overview of Financial Education Evaluations, Consumer Research 68, Prepared for the Financial Services Authority by Personal Finance Research Centre, University of Bristol, July. _____ (2008b) Financial Capability: A Behavioural Economics Perspective, Consumer Research 69, Prepared for the Financial Services Authority by the London School of Economics, July. _____ (2008c) Business Plan 2008/9, February. _____ (2009) Building Financial Capability in the UK. Available from: http://www.fsa.gov.uk/financial_capability/ [Accessed date: 22 January 2009]. Fluch, M. (2007) Selected Central Banks’ Economic and Financial Literacy Programs. In: Monetary Policy & the Economy Q3/07, 85-103, Oesterreichische Nationalbank, Vienna. Gnan, E., Silgoner, M.A., and Weber, B. (2007) Economic and Financial Education: Concepts, Goals and Measurement. In: Monetary Policy & the Economy Q3/07, 28-49, Oesterreichische Nationalbank, Vienna. HM Treasury and Financial Services Authority (FSA) (2008) Helping you make the most of your money: a joint action plan for financial capability, July. Hood, C. (1983) The Tools of Government. Macmillan, London. Hood, C, and Margetts (2006) The Tools of Government in the Digital Age. Palgrave, London. National Audit Office (2007) Government on the Internet: Progress in Delivering Information and Services Online, Report by the Comptroller and Auditor General, July 13. Organisation for Economic Co-operation and Development (OECD) (2005a) Recommendation on Principles and Good Practices for Financial Education and Awareness. Recommendation of the Council., July, Paris. _____ (2005b) Improving Financial Literacy: Analysis of Issues and Policies, Paris. _____ (2009) International Gateway for Financial Education. Available from: http://www.oecd.org/pages/0,3417,en_39665975_39666038_1_1_1_1_1,00.html [Accessed date: 22 January 2009]. U.S. Financial Literacy and Education Commission (2009) MyMoney.gov. Available from: http://www.mymoney.gov/ [Accessed date: 22 January 2009]. 218 IADIS International Conference e-Commerce 2009 A CUSTOMER FOCUSED E-COMMERCE APPROACH USING PURCHASING WEB PATTERNS Markus Weinmann, Yvonne Gaedke Technische Universität Carolo-Wilhelmina zu Braunschweig, Institut für Wirtschaftsinformatik, Abteilung Informationsmanagement, Mühlenpfordtstrasse. 23, 38106 Braunschweig ABSTRACT Why do some users leave web shops during their buying process - and others do not? In this paper, it is assumed that this behaviour depends on the "electronic readiness" of people and on specific situations of customers during the buying process. This paper will describe the approach to analyse barriers to e-commerce adoption by clustering customers according to their e-commerce readiness and by taking into consideration individual situations. The idea is to develop purchasing web patterns to support customers during the e-commerce process. KEYWORDS e-commerce readiness, psychology, web pattern, purchase process, customer focused, situation 1. MOTIVATION The e-commerce revenue in Germany of the year 2008 is 19 % higher than in the year 2007. The study in 2008 with the GFK WebScope-panel shows that online consumers differ within their online buying behaviour [GFK 2009]. What could be done to keep potential customers from getting lost during a buying process? Support is missing and the reusability of support systems is not secured. People differ in their readiness to use the web. There are different aspects and individual situations which influence the unwillingness to buy online. One possibility might be to standardize the e-commerce process. But is this possible? In order to understand the fact that people are leaving the website during a buying process, it is necessary to identify the media behaviour and the situations that occur during that process. Consequentially it will be possible to develop purchasing web patterns which could be customized for the different types of potential customers. 2. E-COMMERCE READINESS INDEX (ECRI) The technology provided for e-commerce processes is already developed - but are the people ready for e-commerce? The first step in this project should be to find an index which explains the level of people’s electronic-commerce readiness. For software engineering it is already common sense to distinguish between experienced and less experienced users. Non experienced users need more assistance, however the professional users are more function orientated [Löffler 1989]. A study of Forrester (2007) among 4000 US citizens regarding the readiness of people to use web 2.0 tools classified the more experienced users in five types: the "viewers", who participate passively and do not create content; the "participators", this group uses social networks; the "assemblers" use RSS and social bookmarks; the "detractors" commentate and write product descriptions; the "makers" create the content. According to Charlene Li (2007), it is possible to motivate people to get into the next user category [quoted by Ecin 2007]. So it is important to understand how people will change their user behaviour and make them ready for e-commerce. 219 ISBN: 978-972-8924-89-8 © 2009 IADIS 2.1 Literature Review Although most of the acceptance and behavioural theories are focused on one specific application or innovation (e.g. mobile banking), for the e-commerce readiness research it is assumed that these models are also alienable to e-commerce processes in general. The e-commerce process is understood as a new technology although it is not an innovation. Before being able to define an index describing the level of users’ readiness to purchase online, it is important to review existing literature. Molla and Licker (2005) summarized the literature and introduced five dominant perspectives relevant to the e-readiness research: the technological imperative perspective, the managerial imperative perspective, the organizational imperative perspective, the environmental imperative perspective and the interactionism perspective. The technological imperative models, such as the diffusion of innovation (DOI), the technology acceptance model (TAM) and the unified theory of acceptance and use of technology model (UTAUT) meditate the complexity, compatibility, relative advantage, task conformity, trust, experiences, self-efficacy, ease of use and usefulness as key variables to adopt a special innovation [Molla and Licker 2005, Venkatesh et al. 2003]. The managerial imperative model examines the innovation adoption by the management, it looks at the innovativeness, the commitment and the IT background of managers. The organizational imperative models survey the structure of a company. There might be different behaviours and attitudes due to functional differentiation, risks, innovativeness or the like. Environmental imperative models take into consideration external influences, which cannot be changed and are important to decision makers. Why some innovations are successful within a certain company and some are not could be explained by the interactionism model, which provides the base for the perceived e-readiness model (PERM). Hereby the key indicators are the perceived organizational e-readiness (POER) and the perceived external e-readiness (PEER), which determine the e-commerce adoption [Molla and Licker 2005]. Lai et al. (2008) found four key constructs to measure the employee readiness for e-business (EREB). These are: benefit, security, collaboration and certainty [Lai et al. 2008]. The literature review shows that there are several different influence factors for adopting new technology and using the technology. This project’s aim is to classify user groups by their e-commerce readiness level and then to develop a model explaining e-commerce adoption and to find the e-commerce readiness index (ECRI) of people. It is assumed that the ECRI is in the online context an additional variable which influences the intention to buy online and the actual purchasing behaviour as it has the perceived control in the theory of planned behaviour [Ajzen 1991]. It is assumed that people with a low ECRI have a low intention to buy online, whereas people with a high ECRI should have a strong intention to purchase online. To proof this coherency it is important to find a valid ECRI. As the ECRI should apply to everyone and not only to employees as the EREB does, it will be modified by further factors taken from the five dominant perspectives. In addition to these factors the personal motivation and selfefficacy are also expected to have an influence on the ECRI. It is supposed that there will be three ECRI clusters: the professionals who have the highest ECRI, the advanced users and the beginners with the lowest ECRI (see Fig. 1). ECRI: Pro ECRI: Advanced ECRI: Beginner Figure 1. Clustering Customers Due to their E-commerce Readiness Index (ECRI) 2.2 Approach and Methods To identify relevant factors for this study, a focus group including university professors, PhD students, students and experts will be interviewed and asked to modify the proposed item list. According to that modified list a questionnaire will be developed. To guarantee representative results the sample N>300 is aspired. The survey will be sent both online and offline. This decision is based on advantages and disadvantages of an online questionnaire mentioned by several researchers [Sarodnick and Brau 2006]. 220 IADIS International Conference e-Commerce 2009 The advantages of an online questionnaire are cost-efficiency, a lower error ratio, the wider group of potential participants and anonymity. One of the most relevant disadvantages is that computer experienced people are more familiar with an online survey than less experienced. This might have an impact on the results as the risk to reach only experienced people, which might be e-readier than others, is high. Therefore there will be also the possibility to answer the questions offline. A cluster analysis will be conducted to classify people according to their ECRI. To verify that the ECRI has an influence on the intention to buy and on the actual online purchase behaviour, potential users will be asked to complete a short questionnaire to define their cluster affiliation and then they will be confronted with a real online purchasing situation in a usability laboratory. It is supposed that there are more factors which have an influence on the actual online buying behaviour. Thus based on the behavioural and acceptance models, a causal model will be developed which can explain the factors that have an influence on purchasing online. One of the factors will be the ECRI. But even if people are familiar with the internet and e-commerce, it can happen that they quit the purchasing process. Users might be in a bad mood or in a hurry or stressed by something that deters them from finishing the purchasing process. So in chapter 3 other factors such as emotional conditions and situations of people will be taken into consideration and a possibility to support people during the e-commerce process will be introduced. 3. PURCHASING WEB PATTERNS To assure that the right reasons why people interrupt the purchasing process will be identified, the customers’ emotional conditions and situations have to be characterized [Belk 1975]. Once they are characterized, it is possible to classify people’s behaviour in different situations. With this information it should be possible to help and support customers during different buying stages. The aim is to detect certain behaviour patterns of people during a purchasing process or to classify them according to their behaviour in order to be able to introduce "purchasing web patterns". These patterns identify the customer’s situation during the purchasing process and support him with the perfect pattern depending on his individual current situation. The main issue of the purchasing web patterns is the reusability and standardization. 3.1 Literature Review Before defining purchasing web patterns, it is important to know the customer’s needs, his choice behaviour and his current situation in order to place them at the right time [Ho et al. 2008, Maslow 1943]. Situations can vary in the different buying stages, which can be adopted form different models like AIDA or DAGMAR (see Fig. 2). These models have to be modified for e-commerce [Colley 1961, Howard and Seth 1961, Strong 1925]. As Woodworth mentioned in his Stimulus-Organism-Response-Model, people are reflecting a stimulus like a situation and react in different ways [Woodworth 1928]. To get the customer involved and to understand his reaction in the different buying stages, it is important to identify the customer’s situation, to give him the right information and to prevent making him dissonant [Berlyne 1975, Bost 1987, Festinger 1957]. Therefore it is necessary to know the actions (response) the customer has to do during the buying process in order to get him to the next step. The action “using the searchmask” may correspond with the situation “being confused”. That means if the response is clearly defined (doing some action) it is fundamental to find the stimulus (situation) which leads to the response. A I D A Situations Figure 2. Situations in Buying Stages The IT system’s task is to observe the customers behaviour in each situation, to interpret it and to help him get to the next action in the buying process if he does not know how to go on. At this point the purchasing web pattern shall be used to bridge situations between two actions. For example if the system 221 ISBN: 978-972-8924-89-8 © 2009 IADIS recognises the users’ situation “being confused”, a purchasing web pattern will be pushed which leads him to the search mask. This support shall be automated by using standardized purchasing web patterns which can be used in similar situations. 3.2 Approach and Methods To develop a model for purchasing web patterns, several steps need to be done. (1) First of all, it is important to identify the different purchase stages in e-commerce context. Therefore a literature review will be made to get an overview over the different models like AIDA and DAGMAR. These models are the starting point to define the purchase stages in e-commerce. (2) The next step is to define tasks within the different purchasing stages. This will also be done with a literature review. (3) Once the tasks in the purchasing process are defined, it is possible to analyse the customer’s situation. That means: Is he able to get to the next task? What are the barriers to achieve the next steps? Therefore an observation in a usability laboratory should be made combined with a self report by the customers. (4) One barrier is assumed to be the purchase pattern of the different customers. A broad literature review combined with self report methods as well as observation in the usability laboratory will be used to get the results. (5) Finally, the customers’ behaviour in different situations during the purchasing process will be analysed and clustered in order to build purchasing web patterns. To improve the capability of the patterns, soft computing techniques such as artificial neural networks and fuzzy logic systems shall be used. 4. COMBINATION AND FORECAST The idea of this customer focused e-commerce research is to help and support him during the buying process. To make sure not to lose the customer during his e-commerce activities the IT system’s task is to prevent him from leaving. The different factors presented in chapter 2 and 3 (a lack of e-readiness, specific situations and emotional conditions of customers) which can stop the buying process will be integrated in one model. The idea is to define in a first step the users’ e-commerce readiness levels and to cluster them into different e-readiness layers in order to get a first indicator how to deal with the customer in certain situations within the whole e-commerce process. In a next step the purchasing web pattern will be defined. It is assumed that besides the e-commerce readiness level the situation and individual emotional state of customers is important during the buying process. So there might be some situations that are always the same in every layer like "start searching" (see Fig. 3 dark points). Some situations might occur only on some levels (see Fig. 3 white points). To help the customer with purchasing web patterns it is necessary to know his e-readiness in order to interpret the right situations. If these factors are known, the customer can be supported during the buying process with the right purchasing web pattern. Not versed customers might need much more support in a lot of different situations then pro users. A I D A ECRI: Pro ECRI: Advanced ECRI: Beginner Figure 3. ECRI-Web-Pattern-Model: Situations in E-commerce-Ready-Index- (ECRI-) Layer During E-commerce Stages A correlation between psychology, people’s situations and the ECRI of people is assumed. To find all causal analytical coherencies, a causal model will be developed and then verified in a usability test situation. During the usability test, test persons are confronted with different purchasing web patterns in different situation during the buying process. The aim of this final usability test is to develop purchasing web patterns, 222 IADIS International Conference e-Commerce 2009 which are able to reduce the abort rate during the whole e-commerce process in order to secure that people do not get lost during the e-commerce process. The ECRI-Web-Pattern-Model is a first step to support customers in the buying process in order not to lose them. But it is not yet completed. There shall be proven whether there are more factors which influence stopping of customers in the e-commerce process. Then the model should be enlarged with these factors. In order to assure that the developed purchasing web patterns support the e-commerce process, they will be proofed continuously in the usability laboratory. REFERENCES Ajzen, I., 1991. The Theory of Planned Behavior. In Behavior and human decision processes, Vol. 50, pp. 179-211. Belk, R 1975. Situational variables and consumer behavior. In Journal of Consumer Research, Vol. 2, pp. 157-164. Berlyne, D.E., 1971. Aesthetics and psychobiology. Meredith Corporation, New York. Bost, E.,1987.Ladenathmosphäre und Konsumverhalten. Physica, Heidelberg. Bandura, A.,1979. Self-efficacy: The exercise of control. Freeman, New York. Colley, R., 1961. Defining Advertising Goals for Measured Advertising Results. Association of National Advertisers, New York. Davis, F.D.,1989. Perceived usefulness, perceived ease of use, and end user acceptance of information technology. In MIS Quaterly. Vol. 13, pp. 310-339. Ecin, 2007. Sechs Sprossen auf der Web 2.0 Leiter. Available at http://www.ecin.de/news/2007/04/26/10663/index.html, accessed on 6.04.2009. Festinger, L, 1957. A theory of cognitive dissonance. Stanford University Press, Stanford, CA. GFK. Without Author, 2009. E-commerce Umsatz wächst weiter. Aktuelle Ergebnisse aus dem GfK WebScope zum Kaufverhalten der Deutschen im Internet. Available at http://www.gfk.com/group/press_information/press_releases/003717/index.de.html, accessed on 6.04.2009. Ho, S.Y., Davern, M.J. and Tam, K.Y., 2008. Personalization and Choice Behavior: The Role of Personality Traits. In The DATA BASE for Advances in Information Systems, Vol. 39, No. 4. Howard, J. A. and Sheth, J. N.,1969. The Theory of Buyer Behavior. Wiley, New York. Lai, J.-Y., Ong, C.-S., Yang, C.-C. and Wang, C.-T., 2008. Assessing and managing employee readiness for embracing ebusiness. In ACM , pp. 79-87. Löffler, L.,1989. Adaptierbare und adaptive Benutzerschnittstellen - Konzeption und exemplarische Realisierung im Bereich der Produktionstechnik. Universität Karlsruhe, Dissertation. Maslow A.H, 1943. A Theory of Human Motivation. In Psychological Review. Vol. 50, pp. 370-396. Molla, A. and Licker, P.S., 2005. eCommerce adoption in developing countries: a model and instrument. In Information and Management. Vol. 42, pp. 877-899. Richard, J., 2008. Internetrecht – Internetkauf. Available at http://www.internetrecht-rostock.de/internetkauf.htm, accessed on 6.04.2009. Sarodnick S. and Brau, H., 2006. Methoden der Usability Evaluation. Wissenschaftliche Grundlagen und praktische Anwendung. Verlag Hans Huber,Bern. Strong, E.K., 1925. Theories of Selling. In Journal of Applied Psychology. Vol. 9, pp. 75-86. Venkatesh, V., Morris, M.G., Davis, G.B. and Davis, F.D., 2003. User acceptance of information technology: Toward a unified view. In MIS Quarterly. Vol. 27, pp. 425-478. Woodworth, R.S., 1928. Dynamic psychology. In Murchison, C. (Ed.): Psychologies of 1925. Clark University Press, Worcester, MA, pp. 111-126. 223 ISBN: 978-972-8924-89-8 © 2009 IADIS ASSESSING THE CONTRIBUTION OF SCM ON E-BUSINESS PERFORMANCE Maria Teresa Borges Tiago, João Pedro Couto and Flávio Tiago University of the Azores, Rua da Mãe de Deus, Ponta Delgada ABSTRACT Supply Chain Management (SCM) continues to gain popularity among companies and has been broadly studied by academic researchers, especially since the last two decades. However, with the development of the digital economy, a new paradigm has emerged in this arena due to the modification occur in supply chain management. Thus, this paper establishes a new model, considering the results of SCM adoption in digital environments on e-business performance, which has been tested in European companies. For that purpose, we used a structural equation modelling (SEM) analysis. KEYWORDS Supply Chain Management, Digital Environments, Performance. 1. INTRODUCTION In the last decades, organizations have begun to realize the importance of closely managing activities of the supply chain in order to create additional value, which can be grounds for significant competitive advantages. Although marketing researchers and information system investigators have studied supply chain management to some extent; there are still few conceptual bases available, necessary to assessing supply chain management’s (SCM) contribution to business success. When analyzed, the on-line performance of the companies and the implications of virtual SCM application, these assessment weakness assume a major role. This paper examines an exploratory survey conducted among a sample of e-business European companies. Using a structural equation analysis, this study explores the relationship between e-business success and SCM initiatives, measured by the internal resources of a company and internal competencies in SCM, and intrinsic success measures, including: sales volume, number of customers, sales area and customer service quality. This work is organized as follows: Section 2 presents the definition and benefits associated with SCM, including virtual supply chain management. An evaluation framework is developed in Section 3 and Section 4 presents the methodology and the achieved results. In the last section, we conclude our study, reiterate major points and suggest avenues for further investigation. 2. LITERATURE REVIEW Traditional supply chain management (SCM) has been widely studied by academics (Oliver & Webber, 1992; Snowdon, 1988). Since the nineties, driven by academic research and organizational practices developed around the concept of e-business, SCM has gained a new dimension and importance. Much research has emerged from the logistics/ distribution and marketing fields as a result, complemented by studies carried out in the information technology field (Nagurney et al., 2002). In light of mounting research, SCM has now acquired the status of a generic management concept that comprised the systemic implementation of processes allowing the development of competitive advantages and profitability of firms through an integrated management of distribution channels (Svensson, 2003). According to the recent research of Wu and Chen (2006), successful supply chain management requires choosing a type of relationship suitable to product and market conditions as well as the adoption of management practices to it. Rapid growth of the Internet as a means for business seems fundamental to 224 IADIS International Conference e-Commerce 2009 reshaping business structure, allowing firms to embrace unprecedented opportunities. In this context, the concept of SCM can be modified into the virtual supply chain management (Apostolou et al., 1999; Gan et al., 2000). While the original value chain model treats information as a supporting element, in a digital era, information itself can be a critical source of value. The introduction of on-line practices has created new opportunities for both suppliers and consumers: (i) firms began to have an open access to larger numbers of suppliers and consumers; and (ii) physical boundaries to consumers were removed (Nagurney et al., 2002; Gabbott & Colgate, 1999). One of the attractive features of Internet is related to the existing three main stages of value-adding informational processes: (1) visibility (improve the ability to track physical operations more effectively); (2) mirroring capability (substitute virtual activities for physical ones and parallel value chain in market spaces); and (3) create new customer relationships (use information matrix to deliver value to customers in new ways). Turner (2000) suggests the classification of virtual supply chain management according to the traditional concept of supply chain management. In this perspective, all the activities of e-business affect differently both the supply and demand side of the value chain. Despite the methodology used to classify different SCM models, academia and managers seem to agree that on-line systems can be used to augment the performance of SCM (Watson et al, 2000; Turner, 2000). In this context, web pages can be considered collaborative tools between stakeholders (Hamel & Sampler, 1998). 3. EVALUATION FRAMEWORK AND HYPOTHESES From the literature review emerges that research developed on supply chain management field focuses mostly on SCM in a physical context, its influence on general business performance and in specific industries. Although academic researchers and practitioners alike praised virtual supply chain adoption (Cooper et al., 1997; Menzer et al., 2001), there remains a lack of empirical evidence as to its effects on ebusiness success. Accordingly, our aim is to establish a measurement framework that helps to fill the current gap in research and provides a better understanding of the critical elements of the virtual supply chain. The framework identifies metrics that can be used to evaluate the impact of the virtual supply chain management on business performance. The proposed research model tests three hypotheses: the first one related to the impact of the virtual supply chain management in firms’ performance and the others regarding the impact of two dimensions in virtual SCM. As our aim is to consider virtual supply chain management effects on e-business performance, we will consider tactical elements. Thus, virtual supply chain management performance will be indirectly measured in this model through two dimensions: communication and integration of partners and planning and control systems. According to the literature, these components profoundly affect the way companies behave in terms of SCM. In terms of e-business performance measurement, the concept in this research is limited to three dimensions (Amit & Zott, 2001): hard factors, soft factors and innovation. The first is an indicator of economic performance, namely number of customers. The second dimension refers to a company’s improvement in customer relationships, measured in our model by the quality of customer service. The last dimension reflects the company’s achievements in terms of its competitive position, given by the sub dimension sales area. Having as reference the achievements of Tan (2002), which noted that all of the significant supply chain management practices have a positive impact on a firm’s performance, hypothesis one is written as: The greater the supply chain management competencies and implementation, the higher a company’s e-business performance on markets. The interactive nature of the Internet allows establishment of this closer relationship, especially by improving the communication among users (Hoffman & Novak, 1996; Watson et al., 2000) and enlarges access to new suppliers and customers (Nagurney et al., 2002). As Eloranta and Hameri (1991) noted, inbound and outbound logistics tend to be separated in research, so in our study we consider them together. Based on this notion the second hypothesis is postulated as: The better communication and integration processes of a firm, the improved performance level of its supply chain management in digital markets. While the Internet is simplifying the nature of communication with and integration of consumers and suppliers, it is also challenging suppliers to find new methods of cost reduction, combined with just-in-time practices (Chopra & Meindl, 2001). As pointed out by Bowersox and Daugherty (1995), Lewis and Talalayevsky (1997) and Van Hoek et al. (1998), the use of technologies of information and communication 225 ISBN: 978-972-8924-89-8 © 2009 IADIS can improve traditional planning and control systems. This leads to the following hypothesis: The better the planning and control systems, the greater the performance level of supply chain management in digital markets. 4. METHODOLOGY AND RESULTS Data from 25 European countries collected by e-Business W@tch annual survey was used to test our three hypotheses. As this study examines primarily the adoption status of virtual supply chain management by companies, it is important to mention that only 360 of the 9,264 total responses were included in this analysis. Distribution of firm size, measured by the number of employees, shows that most cases are small and medium size (around 60% of the firms). The distribution of responding sample is approximately similar to the original sample. The two most heavily represented sectors in the sample are food, beverages and tobacco and transport equipment, with 11.4% and 9.7% respectively, closely followed by all the others. The casual structure proposed in the theoretical framework was examined using a structural equation model. After global model fit had been assessed, numerical results were evaluated in order to test their support of the research hypotheses. The numerical results for our model can be obtained directly from the path coefficients of the structural model presented in Figure 1. We refer to standardized coefficients, which account for scale effects and serve as indicators of the relative importance of the variables. Several goodnessof-fit tests were conducted to assess whether the empirical model could explain the observed data, which suggest that our model fits the underlying data quite well. The hypotheses’ paths were all statistically significant. Figure 1. Structural Equation Model Estimation Our findings generally support the conceptual framework presented, even though some of the relationships found were weaker than expected. With regard to H1, results show that virtual supply chain management implementation contributed 34% to the e-business performance construct. This finding empirically supports the concept that e-business performance can be improved by investment in virtual supply chain management systems. Similarly, communication and integration in SCM context and planning and control of inbound and outbound activities contributes 63% and 96% to virtual supply chain management competencies construct. The results demonstrate a higher relative importance to planning and control than would be expected from literature review, because most of the research in this field emphasizes communication and integration elements. With respect to H2 and H3, the results achieved in the model support these hypotheses. 226 IADIS International Conference e-Commerce 2009 5. DISCUSSION AND CONCLUSIONS As literature review demonstrated, there has been little study that examines virtual supply chain management contributions to e-business performance. However, those works developed to study virtual SCM and ebusiness performance were largely confined to specific industries. With this study, we attempt to fill existing research gaps, presenting results from an empirical investigation based on a cross industry survey. The goal of the current study was twofold: (1) to determine whether the implementation of tactical dimensions (communication & integration and planning & control) is positively linked to virtual supply chain management competencies and (2) to identify the nature of the relationship between virtual supply chain management and e-business performance. The results of this effort have generated some interesting findings. First, the data supports our conceptualization for the virtual SCM construct. Within it, both tactical elements have a positive impact on the maximization of virtual supply chain management implementation. Second, these findings allow us to conclude that implementation of a virtual supply chain management system has a positive impact on e-business performance. According to these results, the concept of virtual SCM as an integrated e-business tool that allows a more profitable relation with customers and suppliers is reinforced as is the need for a daily based planning and control emphasized. This study and its findings will be useful for firms intending to emulate the application of virtual supply chain management, giving insights to managers about the influence of different components of virtual SCM in e-business performance. Some useful preliminary insights are produce, leaving however a considerable number of issues for future research, including the possibility of extending the investigation in order to consider the impact of the virtual supply chain management in terms of competitive strategy and operational effectiveness. Similarly, this study could be expanded to compare firm performance in e-business versus nonvirtual business activities. ACKNOWLEDGEMENT This paper is based on data provided by the European Commission and the e-Business W@tch and funding for this work is granted by FCT – CEEApla, Research Center for Applied Economics. REFERENCES Amit, R. and Zott, C. (2001). Value Creation in E-Business, Strategic Management Journal, 22(6), pp.493-520. Apostolou, D., Sakkas, N. and Mentzas, G. (1999). Knowledge Networking in Supply Chains: A Case Study in the Wood/Furniture Sector, Information Knowledge Systems Management, 1 (3-4) (Autumn/Winter), pp. 267-281. Bowersox, D. and Daugherty, P. (1995). Logistics paradigms: the impact of information technology, Journal of Business Logistics, 16 (1), pp. 65–80. Chopra, S. and Meindl, P. (2001). Supply chain management: strategy, planning, and operation, Prentice-Hall: Upper Saddle River, NJ. Cooper, M., Ellram, L., Gardner, J. and Hanks, A. (1997). Meshing multiple alliances, Journal of Business Logistics, 18 (1), pp. 67–89. Eloranta, E. and Hameri, J. (1991). Experiences of different approaches to logistics, Journal of Engineering Cost and Production Economics, 21, pp. 155–169. Gabbott, M. and Colgate, M. (1999). Information technology and relationship marketing: advances, incompatibilities and opportunity, Australasian Marketing Journal, 7 (1), pp. 102-108. Gan, B., Liu, L., Jain, S., Turner, S., Cai, W. and Hsu, W. (2000). Manufacturing supply chain management: distributed supply chain simulation across enterprise boundaries, in Proceedings of the 32nd Conference on Winter Simulation (Orlando, Florida, December 10 - 13, 2000). Hamel, G. and Sampler, J. (1998). The e-corporation: more than just web-based, it's building a new industrial order, Fortune, 23 (December) pp. 80-92. Hoffman, D. and Novak, T. (1996). Marketing in hypermedia computer-mediated environments: conceptual foundations, Journal of Marketing, 60, pp. 50-68. 227 ISBN: 978-972-8924-89-8 © 2009 IADIS Lewis, I. and Talalayevsky, A. (1997). Logistics and information technology: a coordination perspective, Journal of Business Logistics, 18 (1), pp. 141-57. Oliver, K. and Webber, M. (1982). Supply chain management: logistics catches up with strategy, in Logistics: the strategic issues, edited by Christopher, M. (1992), Chapman & Hall: London, pp.63-75. Snowdon, M. (1988). Moving towards a single market, International Journal of Technology Management, 3(6), pp. 64355. Svensson, G. (2003). Holistic and cross-disciplinary deficiencies in the theory generation of supply chain management, Supply Chain Management: An International Journal, 8 (4), pp. 303-316. Tan, K. (2002). Supply chain management: practices, concerns, and performance issues, The Journal of Supply Chain Management, 38 (1), pp. 42-53. Van Hoek, R., Commandeur, H. and Voss, B. (1998). Reconfiguring logistics systems through postponement strategies, Journal of Business Logistics, 19 (1), pp.33–54. Wu, I. and Chen, Y. (2006). A model for exploring the impact of purchasing strategies on user requirements determination of e-SRM, Information & Management, 43 (4) (June), pp. 411-422. 228 IADIS International Conference e-Commerce 2009 INTERORGANIZATIONAL BUSINESS PROCESSES MODELING: CASE OF AN E-PROCUREMENT SYSTEM Khoutir Bouchbout, Zaia Alimazighi Computer Science Department, USTHB, Algiers, Algeria ABSTRACT Modeling and managing business processes that span multiple organizations involves new challenges, mainly regarding the ability to cope with change, decentralization, and the required support for interoperability. A key to maintain competitiveness is the ability of an enterprise to describe, standardize, and adapt the way it interacts with suppliers, partners, competitors, and customers. In the context of process orientation, enterprises today describe these procedures and interactions in terms of business processes, and invest huge efforts to describe and standardize these processes. The main objective of this paper is to contribute to the understanding of the Business Process Modelling field focusing on the definition of the Inter-Organizational Business Processes (IOBP) from a high-level modelling perspective. In order to clarify all the fundamental concepts that surround this field in an e-procurement system, we investigate commonly used business process languages on their suitability to model IOBP. KEYWORDS Inter-organizational Business Processes, BPMN, UML Activity Diagrams, e-Procurement 1. INTRODUCTION The fast and increasing development of networked business environments brings new ways of interaction among the enterprises, which eliminated the time and space gap between business partners. Interorganizational systems are a new organizational structure that accomplishes the requirements of dynamism and agility that electronic commerce entails. In order to overcome the shortcomings of the traditional document-centric focus, a lot of research has been undertaken to move towards process-centric B2B approaches. Business process models capture the business information that is required in each step of an IOBP. A business process is defined as an organized group of related activities that together create customer value [7]. IOBP is a special type of a business process that involves two or more partners. In this work we examine the concept of IOBP. Then, we analyzed the particularities of and the requirements for IOBP modeling. We focused on recent studies with more intensive forms of IOBP modeling in an e-procurement system. It is therefore important to consider the relevant requirements of IOS and understand the related influences when we model their business processes. The remainder of the paper is structured as follows. First, we present a development method for designing an e-procurement system. Section 3 investigates the field of IOBP modeling approaches and techniques. Finally, section 4 summarized the main ideas of this work and gives an overview of future work. 2. THE E-PROCUREMENT SYSTEM DEVELOPMENT METHOD e-Procurement system in essence, mirror the procurement process through the provision of two distinct, but connected infrastructures, internal processing (corporate intranet) and external communication processing (Internet based platforms). e-Procurement also enlarges customer base, broadens the search for raw materials to lower its production cost. Purchasing goods and services in an organization can be viewed as two processes: supplier selection and the ongoing purchase of materials. The supplier selection process comprises defining the requirements for the purchased items, identifying potential suppliers, requesting supplier quotes, evaluating bids, and ultimately selecting suppliers based on expected performance as well as price. Once 229 ISBN: 978-972-8924-89-8 © 2009 IADIS suppliers are selected, the purchase of materials typically includes computing material requirements on an ongoing basis and then placing orders [4]. Automating the exchange of business information between business partners exists for a while. The business processes between two different organizations participating in a IOBP must be defined. For this purpose a commonly accepted methodology is needed based on the Open-EDI reference model [13]. What is needed are end-to-end, strategy-to-code methods for designing and deploying e-procurement initiatives as illustrated in figure 1. The business strategy phase of e-procurement development identifies “what” to do. The business process engineering phase addresses the “how.” The task at hand is the mapping and engineering of the IOBP. IOBP may be loosely coupled as in the case of open markets. Or they may be tightly coupled as in the case of a supply chain management (e-procurement) where process handoffs are made electronically. The technical layer (technology standards and infrastructure) corresponds to the Functional Service View (FSV) of the Open-EDI model which separates the “what” in the Business operational View (BOV) from the “how” in the FSV, refined into the deployment artifacts layer and the software environments layer. Deployment artifacts comprise business process specifications, workflow descriptions or document schemes in a machine-processable language. The software environments layer corresponds to concrete implementations of information systems. Software environments consume deployment artifacts in order to execute or participate in a certain process. Business Layer e-procurement business strategy Strenghts, weaknesses, opportunities, threats GOALS, REQUIREMENTS, CONSTRAINTS Process Layer Interorganizational business processes Internal and external BUSINESS PROCESS MODELS (EPC, UML, BPMN, UMM) Technology standard & infrastructure Technical Layer Web Services, Workflows, BPEL, XML, RosettaNet, ebXML Figure 1. e-Procurement Systems Development Method 3. INTERORGANIZATIONAL BUSINESS PROCESSES MODELING Over the last years, a lot of methodologies for modeling business processes have been developed. Some of them are based on special notations often defined by standardization bodies. Others customize the UML for business process modeling needs. Traditionally, business process modeling focuses on modeling business processes internal to an organization fulfilling customer needs. More recent approaches also take IOBP into account. Another criteria for distinguishing business process modeling approaches is its binding to the supporting IT infrastructure. Some approaches are mainly used in the requirements specification phase to support the communication with business domain experts. Resulting models usually hide implementation complexity and are on a rather abstract level. Other approaches are more implementation oriented and rather provide a graphical interface for workflow languages or Web Service orchestrations/choreographies. 230 IADIS International Conference e-Commerce 2009 3.1 Requirements for Interorganizational Business Process Modeling The defining characteristic of an IOBP is that two or more autonomous organizations jointly execute a process with the purpose of creating a certain output. Usually, organizational boundaries are associated with a lack of transparency, redundancies (e.g. the manual re-entry of data) and time lags, thereby delaying the process flow. Although most of these inefficiencies are also present in the case of cross-functional coordination, some specific challenges exist at the boundaries of organizations which necessitate deduction of requirements for their representation in future process architectures (see Table I). Table 1. Challenges and Requirements of IOBP Modeling Challenges & requirements External processes as “black box” [8] ¾ Representation of inter organizational business process Lacking clarification of responsibilities at company boundaries ([1], [8]) ¾ Allocation of tasks to actors Different process logic and terminology ([6], [8]) ¾ Alignment of semantics underlying the business process Process autonomy ([2], [8], [11], [18]) ¾ Decoupling of internal and external processes Confidentiality [11] ¾ (Selective) visibility of internal processes to external partners Contractual relationships [8] ¾ Formal specification of process interfaces Complexity of bilateral agreements([14],[8]) ¾ Support for alignment with multiple partners Contributions from BPM and Workflow Management • Graphical representation of inter-organizational business process • Introduction of artefacts related to organization / roles, messages /business documents • Graphical representation using swimlanes, pools or domains • Organizational / role model to include external parties • Data dictionary, glossary • Information modeling • View concepts • Differentiation between public (or external) business processes and private (or internal) business processes • Abstraction concepts • Partner-specific views • Specification of messages (information flow) • Interface descriptions • Modeling / graphical representation of IO process • Reference processes 3.2 Interorganizational Business Process Modelling Architecture To achieve seamless business processes across enterprise borders the heterogeneity of different terminologies and modelling notations used within the organizations have to be overcome. However, autonomy of the different business partners has to be taken into account meaning that an organization should be able to flexibly participate in business relations. In order to make IOBP work, each involved enterprise has to implement not only its internal processes (private processes), but also its external behavior (public processes). A public process is the definition and execution of a formal message exchange so that messages can be exchanged with other enterprises in a pre-defined sequence and with pre-defined message formats over networks [9]. A public process defines an external message exchange of an organization with its partners according to a message exchange protocol such as EDI and RosettaNet [19]. A private process describes internal executable activities that support the activities of public processes. The public processes of two companies have to match in order to allow IOBP to work. For example, if one company sends a purchase order (PO) then the other company must be able to receive the PO in the format sent over the same network. As depicted in Figure 2, these approaches distinguish between the internal process (private process) and the interorganizational process (collaborative). On private process level, organizations model their internal business processes according to a modelling approach or notation that is most suitable for internal demands independently of the modelling methodologies used by the business partners [18]. 231 ISBN: 978-972-8924-89-8 © 2009 IADIS Inter-organizational business process e.g. BPMN Company A Company B Public business process Public business process Private business process Private business process e.g. EPC e.g. UML AD Workflow e.g. BPEL Workflow Figure 2. Inter-Organizational Business Process Architecture Modelling Layers One possible language for modelling IOBP could be the Business Process Modeling Notation (BPMN) which consolidates ideas from divergent notations into a single standard notation. Examples of notations or methodologies are: UML Activity Diagram ([16],[17]), UMM [21], ebXML Business Process Specification Scheme (BPSS) [12], and EPC [15]. 3.3 Interorganizational Business Process Modeling Languages The origins of process modeling languages are quite diverse, although two dominant approaches can be observed; one based on graphical models, and the other based on rule specifications. The main standards addressing business process modeling considered in this work are outlined below. • EPC – Event-Driven Process Chains (EPCs) are a process-oriented modeling technique proposed by Keller et al. [15]. EPC is a business process modeling language, focusing on control flow dependencies of activities in a business process. It is utilized in the ARchitecture of Integrated Information Systems (ARIS) by Scheer [15] as the central method for the conceptual integration of the functional, organizational, data, and output perspective in information systems design. ARIS is a tool set that supports besides other modeling approaches the EPC approach and is continuously extended to support recent developments in the IT-world. The modeling approach is based on a sequence of events and functions (activities) that constitute a business process. Logical connectors (logical AND, OR and XOR) enable the description of branching actions and conditions for the execution of parallel activities. Extended event-driven process chains support the modeling of resources, data objects, organizational units and services. Further, the linkage between processes is supported. EPCs are considered as a method to be used very easily by domain experts to specify business processes. • UML – The Unified Modeling Language is an object-oriented modelling language from the OMG. It specifies and visualizes models of software systems. It has become the generic modelling standard applicable to any software development project, and knowledge of UML is widespread. Its extensive use has raised numerous application and implementation issues by modelers and vendors. UML 2 was produced to address many of these issues — including business process modeling [17]. In order to describe the behavior of a business process, UML activity diagrams might be used. For specifying interactions between participants on a lower level, one might also utilize sequence diagrams. Many UML approaches studied ([5], [8]) offer means to model collaborations between entities and to derive deployment artifacts thereof. • BPMN – The Business Process Modelling Notation is an open graphical notation standard for business process modelling that was developed by BPMI, now merged with OMG [3]. It is based on wellknown flowcharting notations and thus intuitive for business analysts. The objective is to support process management by both technical users and business users by providing a notation that is intuitive to business users yet able to represent complex process semantics. BPMN defines Business Process Diagrams (BPD) [3] to capture a business process. A BPD describes the flow of a process using flowchart techniques. The modeled process might be either internal to a company (private process) or collaborative if executed between two or more participants (public process). Furthermore, BPMN allows modeling the interface that a private 232 IADIS International Conference e-Commerce 2009 process exposes to its outside world. The interface of such a process defines what message exchanges are required in order to interact with it. Process interfaces are called abstract processes. • UMM – UMM has been developed by UN/CEFACT to analyze and design B2B business processes and to concentrate on business semantics [21]. UMM is a UML based methodology for capturing the requirements in an IOBP. It is independent of the underlying transfer syntax. The overall goal of the UMM methodology is to create a global choreography of the business process. If two business partners interacting with each other each defined their own choreography for the business process the resulting choreographies are unlikely to match. UMM pursues a top down approach by first defining the global choreography from which the local choreographies are derived. Hence it is ensured, that both choreographies are complementary. UMM is built upon the UML meta model and defined as a UML profile [21] e.g. a set of stereotypes, tagged values and OCL constraints. UMM’s modeling of business processes pursues a three-level top-down approach. In the Business Domain View (BDV), inter-organizational and internal business processes are described as high-level use case diagrams. Business partner types are defined as participants in a business process. Processes are complemented by activities, business entities and messages. Finally, the Business Transaction View (BTV) defines the choreography of information exchanges and delineates most of the artifacts specific to interorganizational business process modeling. 3.4 Comparative Study These IOBP languages were compared based on their expressive power, status and support. Expressive power refers to the language’s ability to present different kinds of process constructs, patterns and situations that appear in business processes. UML, EPC, BPMN, and UMM are recommended for e-business modelling. This is no contradiction as each language has its own merits, its own tool support and its own environment, in which it can best be used. A comparison addressing the heterogeneity of business process models can be found in [10]. Since BPMN is defined from scratch as a business process modeling notation, it has a definite business centric approach. While BPMN has a focus on business processes, UML has a focus on software design and therefore the two are not competing notations but are different views on systems. We conclude that the differences in the expressive powers of BPMN and UML AD are not significant. In some situations, BPMN provides better expressive power than UML AD, however. It should also be noted that for both modeling languages, the specification is in some cases too vague and it is thus unclear whether certain patterns can be expressed. Another factor considered was thus tool support: very few tools for generating executable business process code from UML Activity Diagrams were available. The original organization responsible of BPMN also stated that UML has a different viewpoint on business processes, i.e., it focuses more on software design. Although this statement must be taken with caution because it was given by the developers of another standard, we do think that BPMN is more easily accessible for both business and IT users. BPMN is able to present private processes, public processes and collaboration processes. Most BPMN elements can be mapped to execution but some are used purely for informative purposes. BPMN specifies mappings to BPEL only for internal processes, but defines no complementary generation of BPEL artifacts for collaborative processes. According to the BPMN specification, BPSS is considered as a target language for collaborative BPD’s. One of the shortcomings of BPMN is that it lacks formal semantics, and the specifications for certain elements can also be considered inadequate for execution purposes. Additionally, the specification does not include an XML interchange format for BPMN diagrams. For this purpose, OMG has subsequently introduced the Business Process Definition Meta-model (BPDM) specification. Since UMM is defined as a UML profile, a business analyst may use any UML tool to model UMM business collaboration models. As we outlined in the beginning, UMM artifacts are based on a specific subset of UML to capture complex business collaborations. Between these artifacts, a number of dependencies and constraints exist. If a regular UML tool is used for UMM, these rules are not enforced. 4. CONCLUSION The IOBP modeling approaches investigated introduce a representation of the IOBP, which uses either an existing modeling notation or its extensions. Specific artifacts are necessary for describing IOBP, among 233 ISBN: 978-972-8924-89-8 © 2009 IADIS them external organizations, roles or partner types as well as messages, business documents and channels. The most important contribution of existing approaches relates to the decoupling of internal and external business processes. They introduce different views on business processes and distinguish between public (or external) processes and private (or internal) processes. Whereas public processes appear to provide stable interfaces with external partners, private business processes might be subject to change more frequently. Research on specifying, modeling, testing and validating, framework for IOBP for an e-procurement systems is where intensive work is needed. Moreover, for e-procurement to scale to the Internet there is a need for efficient integration with all relevant partners, established a priori or on demand. The need for interoperability in B2B is more pronounced than usual partly because of the way businesses operate, the systems they have, and the difficulties created by systems’ autonomy and heterogeneity. It typically requires the intimate knowledge of the underlying communication protocols, data formats and access interfaces. REFERENCES [1] Aalst, W. M. P. v. d. (2000) Loosly coupled interorganizational workflows: modeling and analyzing workflows crossing organizational boundaries. Information & Management, 37 (2), 67-75. [2] Bergholtz M., Jayaweera P., Johannesson P., and Wohed P. (2002), Process Models and Business Models - A Unified Framework, in Proceedings of ER (Workshops) Springer, Lecture Notes in Computer Science, pp. 364–377. [3] BPMN (2006), Business Process Modeling Notation specification, OMG Final Adopted Specification, OMG, February 2006, URL: http://www.bpmn.org/Documents/OMGFinalAdoptedBPMNl-0Spec2006-02-01.pdf. [4] Chen, M., Meixell M.J. (2003), Web Services enabled procurement in the extended enterprise. Journal of Electronic Commerce Research, Vol. 4, No.4. PP.140-155. [5] Dorn J., Grün C., Werthner H., and Zapletal M. (2007), A Survey of B2B Methodologies and Technologies: From Business Models towards Deployment Artifacts. Proceedings of the 40th Annual Hawaii International Conference on System Sciences (HICSS’07). IEEE Computer Society. [6] Gopal, G. and McMillian, E. (2005) Synchronization: A Cure for Bad Data. Supply Chain Management Review , 9 (4), 58-62. [7] Huemer C., Liegl P., Schuster R., Werthner H., and Zapletal M. (2008), Inter-organizational Systems: From Business Values over Business Processes to Deployment. Proceedings of the 2nd International IEEE Conference on Digital Ecosystems and Technologies (DEST2008). IEEE Computer Society. [8] Legner, C. and Wende, K. (2006) Towards an Excellence Framework for Business Interoperability. Proceedings of 19th Bled eConference eValues. [9] Medjahed, B., Benatallah, B., Bouguettaya, A., Ngu, A., Elmagarmid, A. (2003), Business-to-business interactions issues. The VLDB Journal, Vol. 12: 59–85 pp. 59–85. [10] Mendling, J., Nüttgens, M. and Neumann, G. (2004), A Comparison of XML Interchange Formats for Business Process Modelling. In F. Feltz, A. Oberweis and B. Otjacques, editors, Proceedings of EMISA 2004, LNI 56, pages 129–140. [11] Norta, A. (2007), Exploring Dynamic Inter-Organizational Business Process Collaboration. Dissertation, Technische Universiteit Eindhoven, Eindhoven, Netherlands. [12] OASIS (2006), ebXML Business Process Specification Schema Technical Specification v2.0.4. OASIS Open. [13] Open-edi Reference Model, ISO, 2004, ISO/IEC JTC 1/SC30 ISO Standard 14662, Second Edition. [14] Österle, H. (2004), The Networked Enterprise. In Realtime - A Tribute to Hasso Plattner (Kagermann, H., Ed), pp 151-172, Wiley, Indianapolis. [15] Scheer A.-W., Jost W., Hess H. (2006), Corporate performance management. ARIS in Practice. Springer, Heidelberg. [16] Unified Modeling Language Specification, Object Management Group (OMG), version 1.4.2. http://www.omg.org. [17] Unified Modeling Language Specification, Object Management Group (OMG), version .2.0. http://www.omg.org. [18] Volker Hoyer, Eva Bucherer, Florian Schnabel (2008), collaborative e-business process modeling. Business Process Management Workshops, pp. 185-196. [19] http://www.rosettanet.org [20] http://www.ebxml.org [21] http://www.unece.org/cefact/ 234 Posters IADIS International Conference e-Commerce 2009 WEB AND INFORMATION TECHNOLOGY AS CRITICAL THEMES IN THE CONSUMER BEHAVIOR BASED RESEARCH: A STUDY WITH THEMATIC NETWORKS María Isabel Viedma-del-Jesus, Juan Sánchez-Fernández, Francisco Muñoz-Leiva Department of Marketing and Market Research University of Granada Antonio Gabriel López-Herrera Department of Computer Science and Artificial Intelligence University of Granada ABSTRACT In this paper, a bibliometric and visual study of the research on consumer behavior is presented. A bibliometric map showing the associations between the main concepts in the field are provided from 2004 to 2008. This map provides insight into the structure of the consumer behavior based research, visualize the division of the field into several subfields, and indicate the relationships between these subfields. The results show that some important themes were INTERNET, WEB and INFORMATION-TECHNOLOGY. Experts can use this map to forecast emerging trends in the consumer behavior field and they also enable the novice to become familiar with the field. KEYWORDS Consumer Behavior Research, Information Technology, Web, Thematic Networks, Bibliometric maps, Co-word analysis. 1. INTRODUCTION Consumer behavior has been discussed in depth from the perspective of different fields and investigation areas (such as Cost-Reducing Theory, Psychology, Sociology…), using numerous ideas and concepts, constructions, theories and models. Therefore, because of the interdisciplinary nature of the study on consumer behavior, studies which review publications that directly or indirectly influence the consumer behavior are very necessary. In this sense, the purpose of this investigation is to offer an expeditious perspective of the study of consumer behavior during the period 2004-2008, studying and forecasting the current themes. In order to reach this objective, we will study the last five years of consumer behavior based research using the concept of bibliometric map. It show the associations between the main concepts studied in the field and provide insight into the structure of the consumer behavior field. The map also shows the division of the field into several subfields and indicate the relationships between these subfields. 2. METHODOLOGY The bibliometric map is created by the co-word analysis (Callon et al., 1983; Coulter et al., 1998; Whittaker, 1989). Co-word analysis is a powerful technique for discovering and describing the interactions between different fields in scientific research (Bailon-Moreno et al., 2006). Co-word analysis reduces a space of descriptors (or keywords) to a set of network graphs that effectively illustrate the strongest associations between descriptors. In this study, the process of constructing maps was divided into the six steps proposed by Börner et al. (2003), where 942 papers from ISI Web of Science where downloaded on 10th December 2008 with query: 237 ISBN: 978-972-8924-89-8 © 2009 IADIS TS= ("consumer behavior") OR TS= ("consumer behaviour"), where TS field is a search based on the “Topic”. The papers were studied using CoPalRed (EC3 2006) computer program, which visualizes the networks in a strategic diagram. In a strategic diagram, the abscissa axis is centrality, or the external cohesion index. It represents the most or least central position within the overall network. The ordinate axis is density, or the index of internal cohesion. It represents the conceptual development of the theme. So, quadrant I groups the motor-themes of the discipline, given that it presents strong centrality and high density. The themes of the quadrant II (upper-left) are very specialized, but peripheral in character, while those of the lower-left (quadrant III), with low density and little centrality, mainly represent either emerging or disappearing themes. The themes of the lower-right quadrant (quadrant IV) are the most general basic themes, although with internal development not as high as those of quadrant I. 3. STRATEGIC DIAGRAMS FOR THE PERIOD 2004-2008 To analyze the evolution of the consumer behavior field over the last five years, a strategic diagram for the period 2004-2008 is shown in Figures. The volume of the spheres is proportional to the number of documents corresponding to each theme (a number is used to indicate the papers per theme). In the last five years (2004-2008), the principal themes, in number of papers, were CONSUMERBEHAVIOR (218 papers), WEB (42 papers) and INFORMATION-TECHNOLOGY (20 papers). All of them were located in the upper-right quadrant (with high density and centrality) of the strategic diagram, i.e, they were related externally to concepts applicable to other themes (see Figure 1). The theme WEB appears in the period studied (2004-2008) as an important theme in the consumer behavior based research community. It has become one of the most studied themes. This theme is associated (see Figure 2) to keywords such as online and e-commerce, technology, user-interface and usability. The theme INFORMATION-TECHNOLOGY was other motor-theme (upper-right quadrant) in 2004-2008. Figure 1. Strategic Diagram Figure 2. Associated Keywords to WEB Theme REFERENCES Bailon-Moreno R. et al., (2006). The scientific network of surfactants: Structural analysis. Journal of the American Society for Information Science and Technology, 57(7), 949–60. Börner K. et al., (2003). Visualizing knowledge domains. Annual Review of Information Science and Technology, 37, 179–255. Callon M. et al., (1983). From translations to problematic networks - an introduction to co-word analysis, Social Science Information Sur Les Sciences Socials, 22(2), 191–235. Coulter N. et al., (1998). Software engineering as seen through its research literature: A study in co-word analysis. Journal of the American Society for Information Science, 49(13), 1206–23. EC3: Research Group “Evaluación de la Ciencia y la Comunicación Científica” (2006). CoPalRed (v. 1.0) [Software]. Granada: University of Granada. Available in [http://ec3.ugr.es/]. Whittaker J. (1989). Creativity and conformity in science: Titles, keywords, and co-word analysis. Social Science in Science, 19, 473–96. 238 IADIS International Conference e-Commerce 2009 THE ANTECEDENTS OF USEFULNESS IN EXPERIENCED USERS OF WEB-BASED LEARNING MANAGEMENT SYSTEMS Francisco Muñoz Leiva, Juan Sánchez-Fernández, María Isabel Viedma-del-Jesús, Antonio Gabriel López-Herrera Universidad de Granada ABSTRACT Several researches have focused on the adoption stage of a technological innovation from the users’ viewpoint, obtaining theories or models of behavioural decision, traditionally applied in Social Science. Under the principles of the Technology Acceptance Model (TAM) and its different extensions, this study empirically assesses the main antecedents of perceived usefulness. Data was gathered from 132 experienced users of web-based learning management systems. The findings show that the effect the trust and compatibility with learning style are the two strongest determinants of using a web-based learning management system. KEYWORDS Web-based Learning Management Systems, Determinats, Perceived Usefulness, Compatibility, Trust. 1. THE MAIN ANTECEDENTS OF USEFULNESS OF A SYSTEM Perceived usefulness is defined as “the prospective user’s subjective probability that using a specific application system will increase his or her job performance within an organizational context” (Davis, 1989, p. 985). This belief is a subjective measurement as far as the user is concerned and represents the degree to which the system can be useful when seeking a specific objective. More specifically, this is a multidimensional concept related to increased speed of work, the manner work is done, increased productivity and effectiveness as well as other practical aspects (Featherman and Pavlou, 2003). A particular emphasis have been on the construct perceived usefulness by Davis (1989), Hwang and Yi (2002), Moon and Kim (2001), and Saade and Bahli (2005) among others. Considerable research has been done on the acceptance of new information technology (IT) or systems (IS) but less attention has been paid to the assessment of the factors determining perceived usefulness, despite the fact that this belief has been one of the most important factors in the acceptance of web sites (Moon and Kim, 2001; Chen et al., 2002; Pavlou, 2002; Kourfaris, 2002; Featherman and Pavlou; 2003; Bhattacherjee and Premkumar, 2004; Sánchez-Franco and Rondán, 2005). Furthermore, the present study is based on the hypothesis that perceived usefulness will increase intentions to reuse a web-based content management system supporting learning tasks. The outcomes enable an explanation of the usefulness of these systems as from its antecedents trust, compatibility with learning style, risk to privacy, quality of content and ease of use. The findings allow a web designer or IT specialist to design actions aimed at influencing the acceptance and use of the learning technologies by manipulating controllable external factors, such as the structure and contents or other visual characteristics. 239 ISBN: 978-972-8924-89-8 © 2009 IADIS 2. METHODOLOGY: SAMPLING CHARACTERISTICS AND MEASUREMENT SCALES Given the massive use of innovative web-based systems in support of learning or distance learning (Moodle, Lotus LearningSpace, Illias, WebCT,...), the usefulness of some is questioned and there is a need to understand what the main causes are that determine the success of their adoption. This study is based on a web system specifically developed for students and teachers in a marketing department of a southern European university (WebCiM, available at http://marketing.ugr.es). In the teaching dimension, this system combines contact hours with a made-to-measure design. More specifically, the students registered in any of the departmental subjects are provided with (identified) access to a range of teaching material, information and teacher-student communication, news,.... Using a web-based structured questionnaire, we measured the antecedents of acceptance of this innovative web-based teaching management system after eight weeks of training and familiarization with all the options of the WebCiM platform (December 2005). Table 1. Technical Data of Study (Under Simple Random Sampling Conditions) Students registered at the Faculty of Economic and Business Science 132 valid cases 95% Population Sample size Trust level Accepted error for estimate of proportions, for P=Q=0.5* Field work Type of interview ± 8.53% December 9 - 29, 2005 Web-based questionnaire *For 95% confidence, assuming a simple random sample The measurement model was first evaluated and “fixed” following the generally accepted recommendations (Anderson and Gerbing, 1988; Burt, 1976; McCallum, 1986) before estimation of the structural model. The constructs showed suitable psychometric properties and a certain degree of validity. 3. MAIN FINDINGS To measure the influence of the main antecedents on perceived usefulness, the structural equation model (SEM) of formation of usefulness proposed in section 3 was validated. The software used was LISREL 8.7 and the estimation method, Robust Maximum Likelihood (RML), useful for samples not following normal distribution. The figure 1 shows acceptable overall fit, with the most of statistics inside the values recommended by the literature (Hair et al., 1998). Content quality Compatibility -0,04 (n.s.) 0,32 Perceived usefulness 0,13* Ease of use R2 = 0,21 0,55 0,46 Trust -0,31 R2 = 0,57 0,12 (n.s.) Privacy risk R2 = 0,094 *quasi-significant loadings Indicator Satorra chi-squared RMSEA (90% RMSEA) ECVI (90% ECVI) GFI RGFI NFI NNFI IFI RFI CFI AGFI RAGFI Value 263.97 (P = 0.0019) 0.053 (0.033 ; 0.070) 3.29 (2.95 ; 3.70) 0.70 0.82 0.95 0.99 0.99 0.94 0.99 0.62 0.75 Figure 1. Model Obtained (Standardized Parameters) and Fit Indicators The result of the t test for the parameters obtained by the structural model warns that, for a significance level of 99%, only trust and compatibility give a strong explanation of the usefulness dependent variable. 240 IADIS International Conference e-Commerce 2009 Content quality, risk and ease of use fail to present a significant effect on usefulness, although the last can be considered quasi-significant (with a trust level of 90%), just as the theory suggested. Although not significant, the effect of risk on usefulness was positive, despite what was initially thought. An explanation for this could be that the students had a priori knowledge that the teacher could access their personal data for sending messages (grades, news,...) via intranet or e-mail and the higher the infiltration in these data, slightly higher is perceived usefulness (although not significant). Therefore, for learning management systems, this risk is not an inhibitor of their usefulness. 4. CONCLUSIONS AND RECOMMENDATIONS Analysis of the literature on technological innovations shows some of the effects and antecedents of perceived usefulness delimiting a model of usefulness of a b-learning web innovation. The success of this type of web technology in terms of an increase in perceived usefulness depends on the attributes of the web system controllable by the organization responsible for its use or distribution and, obviously, their intervention in the explanation of the usefulness. This research suggests the need for greater effort in generating trust in the use of this type of platform. The reason for this higher effect of trust is that experienced users perceive that part of the guarantee to be obtained from the usefulness of the web-interface depends on the people behind the platform and their offer, mainly in more routine situations with less uncertainty in Administration to Consumer (A2C) exchanges. The results imply that the way to increase perceived usefulness in a new web site should be by seeking arguments that basically refer to the compatibility with the student’s learning style. Other result shows the low effect of risk of privacy loss on usefulness, unlike what was originally thought. This type of b-learning web systems inherently involves a review of the students’ private and personal data, mainly for contact purposes. This will have a slight non-inhibiting effect on the usefulness of the web system. REFERENCES Bhattacherjee, A., & Premkumar, G. P., 2004. Understanding Changes in Belief and Attitude Toward Information Technology Usage: A Theoretical Model and Longitudinal Test, MIS Quarterly 28(2), 229-254. Burt, R. S., 1976. Interpretational confounding of unobserved variables in structural equation models, Sociological Methods & Research, 5, 3-52. Chen, L., Gillenson, M. L., & Sherrell, D. L., 2002. Enticing online consumers: an extended technology acceptance perspective, Information & Management, 39, 705-719. Davis, F., 1989. Perceived usefulness, perceived ease of use, and user acceptance of Information Technology, MIS Quarterly, 13 (3), 319-340. Davis, F. D., Bagozzi, R. P., & Warshaw, P. R., 1989. User Acceptance of Computer Technology: A Comparison of Two Theoretical Models, Management Science 35 (8), 982-1003. Featherman, M. S., & Pavlou, P. A., 2003. Predicting e-services adoption: A perceived risk facets perspective, International Journal of Human-Computer Studies 59, 451-474. Hair, J.F., Anderson, R. E., Tatham, R. L., & Black, W. C., 1998. Multivariate Data Analysis, (Upper Saddle River, NJ : Prentice Hall). Moon, J.W., & Kim, Y.G., 2001. Extending the TAM for a World-Wide-Web context, Information and Management, 38 (4), 217– 230. Pavlou, P.A., 2002. What Drives Electronic Commerce? A Theory of Planned Behavior Perspective, Academy of Management Annual Meeting, Denver, Colorado. Saade, R., & Bahli, B., 2005. The impact of cognitive absorption on perceived usefulness and perceived use of use in online learning: An extension of the Technology Acceptance Model, Information and Management, 42, 317-327. Sánchez-Franco, M. J., & Rondán, J. L., 2005. Web acceptance and usage model: A comparison between goal-directed and experiential web users, Internet Research 15(1), 21-48. 241 ISBN: 978-972-8924-89-8 © 2009 IADIS APPLICATION OF TERNARY AHP Sylvia Encheva Stord/Haugesund University College Faculty of Technology, Business and Maritime Sciences Bjørnsonsg. 45, 5528 Haugesund, Norway ABSTRACT This work is devoted to application of ternary Analytic Hierarchy Process in the process of choosing the most suitable organization for hosting a computer based system. Several alternatives are considered with respect to a predetermined number of criteria. KEYWORDS Fuzzy logic, decisions, multi-criteria. 1. INTRODUCTION The ternary Analytic Hierarchy Process (AHP), [6] facilitates development of a hierarchical structure of a complex evaluation problem. Subjective judgment errors can be avoided and an increase of the likelihood for obtaining reliable results can be achieved by application of the AHP. The ternary AHP is sufficient with respect to sport games [1]. We propose use of ternary AHP in a different aria. We believe that certain decision processes (definitely not all) can benefit from a simpler rating scale. This will contribute to a higher level of consistency and will save both time and efforts for the decision makers. Decision makers find AHP to be a very useful tool. However, an increase of the number of alternatives and criteria results in a larger amount of pair wise comparisons. The latter is time consuming and thus increases the loads of the decision makers. Binary and ternary AHP have been proposed for solving problems that do not require a larger scale of values representing the intensities of judgments [2] and [11]. The rest of the paper is organized as follows. Related work and supporting theory may be found in Section 2. The decision process is presented in Section 3. The paper ends with a conclusion in Section 4. 2. BACKGROUND AHP [6] employs paired comparisons in order to obtain ratio scales. Both actual measurements and subjective opinions can be used in the process. The decision makers make pairwise comparison of independent alternatives with respect to each criterion and among the involved criteria. The elements aij, i, j = 1,2, ..., n in the obtained matrices satisfy the conditions aij > 0, aij = a-1ji, , aij =1 i, j = 1,2, ..., n. The AHP steps are: The number of levels in an AHP hierarchy can vary greatly according to the need of a particular decision situation. Alternatives carry information of either a quantitative nature or a qualitative nature. For n alternatives only n( n − 1) paired comparisons needed be elicited since reciprocal response data is assumed. 2 The relative importance of criteria and preferences among the alternatives is stated based on pairwise comparisons. The standard rating system employs 9-point scale where equal importance is denoted by 1 and extreme importance is denoted by 9. The upper bound is a result of research in psychology indicating 242 IADIS International Conference e-Commerce 2009 humans' inability to consistently repeat their expressed gradations of preference finer than seven plus or minus two. A priority weight vector for the criteria is obtained via a synthesis process based on the preference scores [7]. Final weights representing the priority ordering of the alternatives are calculated. 3. DECISIONS Decision makers' judgements are consistent if ajk = aik, i, j, k = 1,2, ..., n. In this content consistency means that if a basic amount of row data is available than all other data can be logically deduced from it. Application of eigenvectors leads to a very useful consistency measure called consistency index CI, [6]. It is defined as CI = λ max − n n −1 where n is the order of the comparison matrix and λ max is its maximum eigen value. CI measures the transitivity of a preference that is a part of the pair wise comparisons. A random index RI is the mean CI value of random generated matrices of size n, [6]. A consistency ratio CR is defined as CR = CI and is a measure of comparison between a given matrix and a random generated RI matrix in terms of consistency indexes. The upper bound for an acceptable CR is 0.1. A revision of judgements is required if larger values are obtained. Addition or deletion of alternatives can lead to possible rank reversal [9], [10], and [12]. According to [13] change of local priorities can cause rank reversal before and after an alternative is added or deleted. The distributive mode normalizes alternative scores under each criterion so that they sum to one. This creates a dependency on how well all other alternatives perform and hence the potential for rank reversal. In contrast, the ideal mode preserves rank by dividing the score of each alternative only by the score of the best alternative under each criterion, [5]. Educational institutions are jointed in a collaborative network across organizational boundaries. One of their current tasks is to choose the most suitable institution for hosting new computer based system that will be used by all member institutions. A committee formed by members of the four institutions is making all the decisions needed in the application of the AHP. Figure 1. Sensitivity Analysis Stability of priority ranking is often tested applying sensitivity analysis. The importance of the cost for technical support criterion can be seen in Fig 1. 243 ISBN: 978-972-8924-89-8 © 2009 IADIS Figure 2. Head-to-head Sensitivity Analysis A head-to-head sensitivity analysis between two alternatives shows the relative magnitude of the alternatives compare with respect to the involved criteria, Fig. 2. Both Fig. 1 and Fig. 2 are obtained via the AHPproject - Free Web-Based Decision Support Tool [4]. 4. CONCLUSION The alternative H turns out to be the best choice for a host organization. It is worth noticing that both the distributive mode and the ideal mode show that alternative H has received the highest ranking. The head-tohead analysis was performed for all alternatives but we choose to present graphically a comparison between the first two alternatives only. REFERENCES [1] C. Genest, F. Lapointe, and W. Drury, On a proposal of Jensen for the analysis of ordinal pairwise preferences using Saaty's eigenvector scaling method, Journal of mathematical psychology, 37(4), 1993, pp. 575-610. [2] R. E. Jensen, Comparison of consensus methods for priority ranking problems, Decision Sciences, 17, 1986, pp. 195211 [3] S. Hill and C. Zammet, Identification of community values for regional land use planning and management. International Society of Ecological Economists Congress, Canberra, Australia, 26, 2000. [4] http://www.ahpproject.com/ [5] I. Milleta and T. L. Saaty, On the relativity of relative measures accommodating both rank preservation and rank reversals in the AHP, European Journal of Operational Research, 121(1), 2000, pp. 205-212. [6] T. L. Saaty, The Analytic Hierarchy Process, McGrawHill, NewYork, 1980. [7] T. L. Saaty, Fundamentals of Decision Making and Priority Theory with the Analytic Hierarchy Process, 6, RWS Publications, Pittsburgh, PA, 2000. [8] T. L. Saaty, The Analytic Network Process, RWS Publications, 2001. [9] T. L. Saaty and L. G. Vargas, Experiments on rank preservation and reversal in relative measurement, Mathematical and Computer Modelling, 17(4-5), 1993, pp. 13-18. [10] S. Schenkerman, Avoiding rank reversal in AHP decision-support models, European Journal of Operational Research, 74, 1994, pp. 407-419. [11] I. Takahashi, AHP Applied to Binary and Ternary Comparisons, Journal of Operations Research Society of Japan, 33(3), 1990, pp. 199-206. [12] E. Triantaphyllou, Two new cases of rank reversals when the AHP and some of its additive variants are used that do not occur with the multiplicative AHP, Journal of Multi-Criteria Decision Analysis, 10, 2001, pp. 11-25. [13] J. Wang, Multi-criteria decision-making approach with incomplete certain information based on ternary AHP, Journal of Systems Engineering and Electronics, 17(1), 2000, pp. 109-114. 244 Doctoral Consortium IADIS International Conference e-Commerce 2009 FLOW AND ONLINE CONSUMER BEHAVIOUR: AN EMPIRICAL ANALYSIS OF E-LEARNING EXPERIENCES Irene Esteban-Millat Written under the supervision of Dr Inma Rodríguez-Ardura and Dr Antoni Meseguer Universitat Oberta de Catalunya Doctoral Programme on Information and Knowledge Society-2009 ABSTRACT This research project proposes to advance in the comprehension of online learning experiences in general and, in particular, of the states of flow that emerge in the consumption of higher-education training products in virtual environments. To this purpose, we propose to develop and test empirically a flow model. This initiative aims to clarify the ambiguities and inconsistencies existing in the literature regarding the conceptualisation and operationalisation of flow in general websurfing and consumption contexts, and in turn to characterise the learning process in virtual learning environments. KEYWORDS Flow; learning behaviour; e-commerce; training/learning environments, training products, structural equation modelling. 1. INTRODUCTION In their drive to understand online learning behaviours, researchers have often turned to the approaches habitually used in analysing behaviour in the conventional media, but they have also considered new elements, like flow, which are especially applicable in virtual environments. In spite of the efforts made by scientific research in this line, the literature presents serious discrepancies and inconsistencies regarding the conceptualisation and operationalisation of the concept of flow. In order to shed some light on these issues, various authors have recommended, among other measures, studying them in specific areas of learning. One area of special interest is the consumption of online higher-education training products because of the significant and growing importance acquired in recent years by virtual learning environments in the range of services offered by educational centres. However, little attention has been paid to this, and the results presented by the studies on flow carried out in this field cannot be considered conclusive. 2. REVIEW OF THE LITERATURE Flow (Csikszentmihalyi, 1975) is described as a pleasurable and fulfilling experience in itself (Pierce, 2005). It is considered a complex concept because it is related to a large number of elements of a fundamentally subjective or affective nature (such as the balance between challenges and skills, perception of control, focalised attention, involvement, perception of enjoyment, telepresence, time distortion and interactivity). The concept of flow has been adapted and increasingly used in computer-mediated environments to explain consumers' behaviour when they surf the Net (Hoffman and Novak, 1996; Chen et al., 2000; Novak et al., 2000).2000). The fact is that providing optimal websurfing and online shopping experiences, characteristic of a state of flow, can give rise to desirable results in consumers' consumption behaviour (Novak et al., 2003), such as a favourable attitude towards the brand or a longer duration of the visit to the website, among others. In addition, other positive effects of flow have been confirmed, such as learning (Ghani, 1995) and positive subjective experiences (Hoffman et al., 1996; Novak et al., 2000; Shin, 2006). 247 ISBN: 978-972-8924-89-8 © 2009 IADIS Moreover, and in view of the growing importance of the Net in the sphere of university education, the study of the concept of flow is especially interesting in the specific context of the consumption of highereducation training products. Furthermore, since these institutions are coming under increasing pressure from market forces and other factors of the environment, which leads them to adopt principles and strategic orientations that focus the organisation's efforts on the satisfaction of the user or consumer (in this case, the student) and the establishment of close, prolonged and mutually beneficial relationships (Daradoumis et al., 2008). As a result, detailed study of the behaviour of their end consumers takes on special importance. To date, approaches in this field via the concept of flow have been based mainly on the identification and/or analysis of some of the factors that facilitate the flow experience (Pearce, 2005; Pearce et al., 2005; Shin, 2006; O'Broin and Clarke, 2006; and Choi et al., 2007). However, there is a shortage of consistent flow models that can be applied to training environments and that will help to respond to many questions for which we do not yet have definitive answers (Pierce, 2005). 3. RESEARCH QUESTIONS AND OBJECTIVES Realising the importance of more extensive knowledge of online learning experiences via the concept of flow and its application in higher-education e-learning environments, we propose a research task with the following primary objectives: PRIMARY OBJECTIVE 1: To confirm the existence of forms of flow in e-learning experiences. In spite of the scarcity of research work carried out in the sphere of consumption of online training products and the fact that the results obtained by them cannot be considered conclusive (they are based on small-sized or convenience samples, etc.), they do indicate the existence of flow sensations in e-learning experiences. It is therefore proposed to contribute empirical evidence that will confirm the existence of flow phenomena during websurfing in virtual learning environments and online consumption of training products. Consequently, we put forward the following research questions: Research Question 1.1: Do forms of flow exist in the sphere of e-learning? PRIMARY OBJECTIVE 2: To identify the flow antecedents and the interaction between them. Although there is a certain degree of consensus among researchers regarding the definition of flow, there seems to be some disagreement, or at least a certain ambiguity, in specifying what elements are its antecedents (see Novak et al., 2000; and Chen et al., 2000; among others). In addition, it would be useful to analyse the relationship of other variables, not customarily contemplated by the literature on flow, with this concept. For instance, it would be helpful to test what elements of website design (user interface, surfing structures, etc.) facilitate flow experiences, and how they do so. It would also be useful to consider the relationship of flow with individual differences (psychological, demographic and socio-economic) between consumers, their experience in the Net, situational factors and the connection quality of the line, among others. To this purpose, an answer is sought to the following research question: Research Question 2.1: What are the flow antecedents in e-learning environments? Likewise, there does not appear to be a consensus among researchers concerning the interactions existing between certain antecedents of the concept of flow. And, although the model developed by Novak et al. (2000) to explain some of these interactions is notable for its exhaustiveness and importance, it seems necessary to study these aspects in greater detail, both to identify new relationships and to detect the causal links between them. For this reason, the following research question is posed: Research Question 2.2: What are the interactions between flow antecedents in e-learning environments? PRIMARY OBJECTIVE 3: To identify the main consequences of the state of flow. In the literature, various consequences of the state of flow are proposed, most notably, due to being the most frequently mentioned, an increase in the duration and repetition of visits (Webster et al., 1993; Shih, 1998; and Koufaris, 2002). However, it has only been possible to validate some of them, such as positive subjective experiences and learning of websurfing techniques, as direct consequences of the state of flow. Hence the interest in testing empirically some of the attitudinal and behavioural consequences of flow proposed by previous studies. Accordingly, the following research question is posed: Research Question 3.1: What are the main consequences deriving from the flow experience in e-learning environments? In the specific field of e-learning, it would be interesting to probe more deeply into the specific study of one of the possible consequences of flow, namely learning. Several works suggest a positive relationship 248 IADIS International Conference e-Commerce 2009 between flow and learning (Csikszentmihalyi, 1989; Ghani et al., 1991; Webster et al., 1993). It should be taken into account, however, that the relationship between the two elements is not simple, and neither are the measurement of flow nor the specification of the nature of the learning process (Pearce et al., 2005). In this way, it is proposed to answer the following research question: Research Question 3.2: How do learning processes deriving from the state of flow in e-learning environments develop? PRIMARY OBJECTIVE 4: To determine the relationship between flow and learning behaviours. The literature presents contradictory results on the type of orientation in websurfing that best facilitates the state of flow: whether it is the type whereby the consumer aims to satisfy clear consumption goals, guided by rational impulses that lead him/her to search for the best option; or another more spontaneous and exploratory type that manifests itself in the search for pleasant consumption experiences. While some studies (Novak et al., 2003) show that websurfing strategies oriented to achieving particular consumption goals are those that best facilitate the state of flow, others (Hoffman and Novak, 1996; and Novak et al., 2000), in contrast, indicate that the adoption of a hedonic strategy better facilitates the emergence of flow experiences. This makes it necessary to conduct further research to provide clarifying answers. This is the motive underlying the following research question: Research Question 4.1: What is the relationship between flow and the orientations of online behaviour in e-learning environments? In accordance with the proposal formulated by some authors (Novak et al., 2003; among others) to shed light on this question, this study proposes to confirm the existence of two differentiated forms of flow, each one of which would be predominantly linked to a specific websurfing strategy or orientation. In this way, the aim is to affirm the existence of possible differences in determining the state of flow according to the type of websurfing (oriented to clear consumption or hedonic goals) carried out by the consumer. The study will also seek to evaluate, for each defined form of flow, the antecedents that determine its emergence and also the perceptions, attitudes and behaviours that are specific consequences of it. In addition, it would be interesting to identify what type of flow best facilitates the development of desirable consumption behaviours for companies (such as favourable attitudes towards the brand and the online supply, longer duration of visits or intention of repeating the visit in the future). This could be the basis for an answer to the following questions: Research Question 4.1: Are there e-learning behaviours that occur predominantly in certain forms of flow? Research Question 4.2: What form of flow best facilitates desirable e-learning behaviours for the organisation? In the event of confirming the existence of two forms of flow, each linked to a specific websurfing orientation, it could also be verified whether these can occur at the same time in one and the same learning experience. In addition, and in the event that two forms of flow can indeed occur in the same learning experience or situation, it will be useful to determine whether one of the two dominates over the other and contributes more to the learning experience. In this way, the study seeks to answer the following research question: Research Question 4.3: Does one form of flow dominate over another in the online learning experience? 4. METHODOLOGY In view of the lack of consensus on many constructs related to flow and the scarcity of research into this phenomenon in e-learning environments, it was considered advisable to carry out an initial qualitative analysis on the basis of two group dynamics. The expectation is that the results deriving from this analysis, along with the evidence provided by the literature, will make it possible to draw up a proposal of flow-related elements in e-learning experiences. The next step consists in defining the items which are to measure each of the elements proposed by means of Likert (multi-item) scales.Preceded by exploratory factor analysis, the structure of the scales will be assessed for reliability through a SEM-Lisrel application. After configuring a complete analytical flow model in the field of online training products, it will be empirically tested by applying the following steps: - Summary of key features of the conceptual flow model. 249 ISBN: 978-972-8924-89-8 © 2009 IADIS - Definitions and operationalisation of constructs (application of measurement scales). - Formulation of hypotheses to be tested empirically, deriving from the research questions. - Development of a quantitative survey instrument (preceded by a pilot test in order to check its coherence and assess the validity and reliability of the measurement scales of the constructs included). - Data collection. The population will consist of students of the Open University of Catalonia (UOC) with experience in e-learning, from which a minimum sample of 400 individuals will be selected. The instrument will be administered in an electronic format, via an email message including a link giving direct access to the website questionnaire. - Analysis of results. Through the use of a structural modelling approach we will test the conceptual model: firstly by purifying the measurement model and secondly by fitting the base model. 5. WORK DONE SO FAR Review of the literature in the following topics or lines of research: consumers' behaviour and its determining factors, online consumption motivations and behaviour, online flow experiences, marketing-education interface, state of flow in e-learning. REFERENCES Chen, H. et al., 2000. “Exploring web users’ optimal flow experiences”, Information Technology & People, Vol. 3 (4), pp. 263-281. Choi, D.H. et al., 2007. “ERP training with a web-based electronic learning system: the flow theory perspective”, International Journal of Human-Computer Studies, Vol. 65 (3), pp. 223-243. Csikszentmihalyi, M., 1990. Flow: the psychology of optimal experience. New York: Harper and Row. Daradoumis, T. et al. (Forthcoming). “CRM and higher education: developing a monitoring system to improve relationships in e-learning environments”, International Journal of Services Technology and Management. Ghani, J.A. et al., 1991. “The experience of flow in computer-mediated and in face-to-face groups”, in J.I. Degross; I. Benbasat; G. Desanctis; C.M. Beath (eds.), Proceedings of the Twelfth International Conference on Information Systems, pp. 229-237. New York: ICIS. Hoffman, D.L.; Novak, T.P., 1996. “Marketing in hypermedia computer mediated environments: conceptual foundations”, Journal of Marketing, Vol. 60 (3), pp. 50-68. Koufaris, M., 2002. “Applying the technology acceptance model and flow theory to online consumer behaviour”, Information Systems Research, Vol. 3 (2), pp. 205-223. Neville, K. et al., 2002. “Mentoring distance learners: an action research study”, in X European Conference on Information Systems. Gdañsk (Poland): Universität Trier. Novak, T.P. et al., 2003. “The influence of goal directed and experiential activities on online flow experiences”, Journal of Consumer Psychology, Vol. 3 (1/2), pp. 3-16. Novak, T. P. et al., 2000. “Measuring the customer experience in online environments: a structural modeling approach”, Marketing Science, Vol. 19 (1), pp. 22-42. Pearce, J. et al., 2005. “The ebb and flow of online learning”, Computers in Human Behavior, Vol. 21 (5), pp. 745-771. Pearce, J., 2005. “Engaging the learner: how can the flow experience support e-learning?”, in G. Richards (ed.), Proceedings of World Conference on E-Learning in Corporate, Government, Healthcare, and Higher Education, pp. 2.288-2.295. Chesapeake (Virginia): AACE. Rodríguez-Ardura, I., 2008. Marketing.com y comercio electrónico en la sociedad de la información (3ª. ed.). Madrid: Ediciones Pirámide and ESIC Editorial. Shin, N., 2006. “Online learner’s flow experience: an empirical study”, British Journal of Educational Technology, Vol. 37 (5), pp. 705-720. Webster, J. et al., 1993. “The dimensionality and correlates of flow in human computer interactions”, Computers in Human Behavior, Vol. 9 (4), pp. 411 -426. 250 IADIS International Conference e-Commerce 2009 THE IMPACT OF CULTURAL ADAPTATION ON THE EFFECTIVENESS OF E-COMMERCE WEBSITES Femke Vyncke, Malaika Brengman, Olga De Troyer Vrije Universiteit Brussel Pleinlaan 2, 1050 Brussels, Belgium ABSTRACT In an online business context, cultural adaptation (Singh and Matsuo, 2004) or localization (Cyr and Trevor-Smith, 2004; Okazaki, 2004) is the intensive process of designing or adapting a website to accommodate the culture-specific needs of online customers in host markets. The current doctoral research aims to investigate whether cultural adaptation improves the effectiveness of e-commerce websites and which website features are important in this matter. This study, as such, aspires to (1) deliver a substantial contribution to the standardization/adaptation debate following the recommendations made by Taylor (2005) and (2) meet the need for systematic cross-cultural studies on the effectiveness of website design aspects appealed for by Cyr (2008). KEYWORDS Culture, website design, effectiveness. 1. PROBLEM STATEMENT AND RESEARCH OBJECTIVE The main problem statement of this research tackles the question of whether e-commerce websites need to be localized in order to attract and retain international clients and, if this is the case, which website features would require localization. Localization, in an online context, is the intensive process of adapting a website to a particular language, cultural expectations, and desired local “look-and-feel” (Cyr and Trevor-Smith, 2004; Singh and Baack, 2004; Okazaki, 2004; Steenkamp and Geyskens, 2006). It entails creating country-specific, culturally congruent websites. In localizing a website such details as local terminology, tone of the text, time zones, currency, country specific symbols, local color sensitivities, navigational preferences, gender roles, and geographic examples must all be considered. Localization is also referred to as cultural adaptation (Singh and Matsuo, 2004) or local adaptation (Taylor, 2005). To investigate how cultural adaptation impacts the effectiveness of transaction-oriented e-commerce websites, a comprehensive experimental design will be set up. In this experiment, a systematic refined approach will be applied, in which internet users from different EU countries will be confronted with fictional websites that will be adjusted to several cultural dimensions on different domains (different website features) and at various levels (low or high reflection of a certain cultural value or dimension). The effectiveness of the different websites will be measured using an extensive online survey. This way the dissertation aims to find out to which extent cultural adaptation is desired and to which degree localization has an impact on the different effectiveness measures. 2. STANDARDIZATION OR ADAPTATION? Whether companies should implement global marketing strategies, and especially how, has turned out to be a complex matter that has yet to be solved in the academic world (Usunier and Lee, 2005). The standardization vs. adaptation debate, in which it is investigated whether marketing stimuli need to be adapted to cultural differences between countries, has been taking place for a while for offline marketing communication (i.e. 251 ISBN: 978-972-8924-89-8 © 2009 IADIS print and television advertising, please refer to Theodosiou and Leonidou, 2003; De Mooij, 2004 and Taylor, 2005) but has only just taken off in the domain of online communication (Okazaki, 2004; Steenkamp and Geyskens, 2006; Brengman, 2006; Baack and Singh, 2007; Cyr, 2008). Two main characteristics of the World Wide Web are of interest in this debate. On the one hand the internet is a medium of mass communication, accessible to everyone, which speaks in favor of standardization. On the other hand it is a very interactive medium, which enables companies to retrieve information on internet segments on the basis of which a high level of adaptation to the user can be introduced. Within the ‘bricks-and-mortar’ retailing context it has been noticed that whereas standardization strategies have successfully been implemented in the past 25 years, the necessity to adapt to local needs and wishes (= localization) has come forth recently (Rigby and Vishwanath, 2006). The question is then whether, within a ‘bricks-and-clicks’ or ‘pure’ e-tailing context, localization or standardization is desirable and for which cultural dimensions this needs to be done. 3. CULTURE AND CULTURAL SENSITIVITY OF WEBSITES According to social-scientific approach, culture can be defined as a predisposition in attitude and behavior that is kept in place and transmitted through social upbringing and social intercourse. It is a collective phenomenon of which the core is formed by positive and negative values which, according to Hall (1976), Hofstede (2001) and Schwartz (1994) can be summarized into a limited number of underlying cultural dimensions as can be seen in Table 1. Table 1. Important Cultural Frameworks Author Hall (1976) Hofstede (2001) Schwartz (1994) Cultural dimensions High-context culture vs. low-context culture Masculinity vs. femininity, High vs. low uncertainty avoidance, High vs. low power distance, Collectivism vs. individualism and Long-term vs. short-term orientation Conservatism vs. autonomy (intellectual and affective), Hierarchy vs. egalitarianism and Mastery vs. harmony Since the Internet is a global communication and transaction medium (Steenkamp and Geyskens, 2006), technically one central website of an organization should be sufficient to reach all consumers, everywhere in the world, 24/7. However, looking at online reality suggests that websites are a culturally sensitive medium. Many internationally active organizations operate not one, but several websites, each aimed at a different target country or group of countries (= a different national culture). To illustrate this: on January 14, 2009, Swatch operated websites for 35 different countries, Unicef operated 110 websites and Coca-Cola even as much as 125 websites. A mere glimpse at these websites already reveals differences in type of visuals, menu layouts, etc. The recent academic research that aimed itself at investigating the influence of culture on website design and the need for localization or standardization ended up delivering contradicting results. Several content analysis studies have been carried out in order to investigate whether existing websites on the web are culturally congruent with the country they are targeting (i.e. quantification and scientific exploration of the insight explained in the previous paragraph). Some of these studies conclude that websites are culturally neutral (= standardized) marketing stimuli (Callahan, 2005; De Troyer et al, 2006) whereas others conclude that cultural adaptation indeed takes place in websites (Okazaki and Rivas, 2002; Robbins and Stylianou, 2003; Cyr and Trevor-Smith, 2004; Singh and Matsuo, 2004; Singh et al, 2005; Brengman, 2006). Several studies also investigated the impact of a culturally sensitive website design on website effectiveness (although only very few apply a true experimental design). Here again two groups of results can be distinguished. Some researchers believe that cultural factors have no impact on perceptions of websites and recommend a standardized approach (Hermans and Shanahan, 2002; Dou et al, 2003; Liu et al, 2004). Other studies by Fink and Laupase (2000), Luna et al (2002), Warden et al (2002), Singh et al (2004), Singh et al (2006), and Cyr (2008) confirm, however, that the internet is not a culturally neutral medium and that culturally sensitive design indeed improves websites’ effectiveness. They show that there are intercultural differences in the perceptions and the expectations of website content and emphasize that values, look and feel, themes and symbols in websites need to be adapted to local cultures. 252 IADIS International Conference e-Commerce 2009 4. FILLING THE SCIENTIFIC GAP From the previous paragraph it may be concluded that no conclusive answer to the standardization/localization question with regard to websites can be provided. Additionally, it must be noted that studies so far focused to a large extent on the practice of localization by making comparisons of existing websites across different nations through content analyses (etic approach). Attempts have also been made to study the impact of cultural adaptation on effectiveness by measuring reactions on existing potentially culturally congruent websites (emic approach). Effectiveness studies using true experimental designs are however scarce and those that exist mostly focus on a narrow range of website features (Warden et al, 2002 for example only focus on language). In the proposed study the aim is to examine, in an experimental design, reactions to self-developed and systematically manipulated websites adapted along different cultural dimensions, in order to investigate the level at which cultural adaptation is desirable and the impact the different degrees of localization can have on effectiveness measures. 5. RESEARCH PLAN 5.1 Completed Work First of all a comprehensive conceptual framework had to be developed focusing on the standardization and adaptation of websites and the effectiveness of websites: • The development of this framework has been based on a thorough literature study that has brought a better notion of (1) the effectiveness of communication in an intercultural context and in the adaptation/standardization debate, (2) the influence of cultural values on consumer behavior, more specifically in relation to the use of the internet and specific internet needs, and (3) the ‘localization’ or ‘(inter)cultural adaptation’ of websites and the impact on the website user and the effectiveness of websites. o TIMING: Literature study finished, 1 paper finished but not yet published (‘What do content analytic studies reveal regarding the cultural sensitivity of websites? An overview of empirical evidence according to a vote-counting method’ by Vyncke, Brengman and de Troyer) and completion of a second paper foreseen by end of 2009. • Essential insights were also drawn from explorative research of the practice of adapting websites and the effectiveness of intercultural adaptation versus standardization based on expert opinions and analysis of ‘best practices’ o TIMING: Explorative research has already started (case study within Honda, based on interviews and internal guideline documents concerning international website design), completion foreseen by end of 2009). 5.2 Next Steps For the empirical section of this study the focus will be on the European Union. Creating an overview of the cultural and institutional differences and similarities between countries and regions within the European Union will enable us to identify regions (clusters) between which the desirability of adaptation of ecommerce websites can be examined. Besides cultural differences, national context factors will also be taken into account (e.g., political regime, legal framework, tax climate, specific retail situations and e-readiness, with, amongst others, aspects such as the internet penetration and the internet use) TIMING: Foreseen for first half of 2010. Finally various experiments will be set up to investigate whether Europeans, due to their cultural and/or institutional differences, indeed react differently to websites that reflect different cultural values. The goal is to do this for a number of e-commerce websites, across different sectors. The cultural adaptation of these different experimental websites will be done on the basis of different layers per cultural dimension, so that a full-factorial experimental design can be set up (e.g., high versus low ‘reflected Power distance’ x high versus low ‘reflected Uncertainty avoidance’ x high versus low ‘reflected Masculinity’, etc.). Experimental 253 ISBN: 978-972-8924-89-8 © 2009 IADIS subjects from various European clusters that score specifically low or high for the respective cultural dimensions will be confronted to websites that may or may not be adapted along these cultural dimensions (monadic experiment: 40 respondents per experimental condition, recruited through an internet panel). An online survey will then be used to question their attitude towards the website, their satisfaction, their trust in the source, etc. (using amongst others the WebQual instrument developed by Loiacono et al, 2007), as well as their attitude towards the products/services for sale on the site and their intentions to purchase these items. Additionally, the respondents’ true cultural values will be measured, in order to keep into account the moderating impact of individual and regional differences within the national population (e.g., Schwartz, 1994). The experimental data will be processed and analyzed in SPSS, using multivariate analysis techniques and structural comparison methods (AMOS, Lisrel), in order to give an answer to the question of whether the development of different websites should be done taking into account the cultural differences between regions within the EU and which cultural dimensions are the most important in the process of localization. TIMING: Mid 2010 till beginning of 2011: set up of experiments and website designs. Till mid 2011: execution of experiments. Till beginning 2012: processing of results. Mid 2012: final reporting and doctoral defense. REFERENCES Baack, D. and Singh, N., 2007. Culture and Web Communications. Journal of Business Research, Vol. 60, No. 3, pp. 181-188. Brengman, M., 2006. Culturele Communicatie via het Internet. Een Vergelijking tussen Belgische en Nederlandse Websites, in Duyck, R. and Van Tilborgh, C. (ed.) Management Jaarboek. PIMMS en Vlaamse Management Associatie, Wetteren, Belgium. Callahan, E., 2005. Cultural Similarities and Differences in the Design of University Websites. Journal of ComputerMediated Communication, Vol. 11, No. 1. Cyr, D., 2008. Modeling Website Design across Cultures. Relationships to Trust, Satisfaction and E-loyalty. Journal of Management Information Systems, Vol. 24, No. 4, pp. 47-72. Cyr, D. and Trevor-Smith, H., 2004. Localization of Web Design. An Empirical Comparison of German, Japanese, and United States Web Site Characteristics. Journal of the American Society for Information Science and Technology, Vol. 55, No. 13, pp. 1199–1208. De Mooij, M., 2004. Consumer Behavior and Culture. Consequences for Global Marketing and Advertising. Sage, Thousand Oaks, California. De Troyer, O. et al, 2006. On Cultural Differences in Local Web Interfaces. Journal of Web Engineering, Vol. 5, No. 3, pp. 246-265. Dou, W. et al, 2003. Consumer Patronage of Ethnic Portals. International Marketing Review, Vol. 20, No. 6, pp. 661677. Fink, D. and Laupase, R., 2000. Perceptions of Web Site Design Characteristics. A Malaysian/Australian Comparison. Internet Research, Vol. 10, No. 1, pp. 44-55. Hall, E., 1976. Beyond Culture. Doubleday and Company, Garden City, New York. Hermans, M. and Shanahan, K., 2002. The Reification of Levitt. Advertising Preferences for Mexican and American Online Consumers. Developments in Marketing Science, Vol. 25, pp. 147. Hofstede, G., 2001. Culture’s Consequences. Comparing Values, Behaviors, Institutions and Organizations across Nations, 2nd edition. Sage, Thousand Oaks, California. Liu, C. et al, 2004. American and Taiwanese Perceptions Concerning Privacy, Trust, and Behavioral Intentions in Electronic Commerce. Journal of Global Information Management, Vol. 12, No. 1, pp. 18-40. Loiacono, A. et al, 2007. WebQual. An Instrument for Consumer Evaluation of Web Sites. International Journal of Electronic Commerce, Vol. 11, No. 3, pp. 51-87. Luna, D. et al, 2002. Cross-Cultural and Cognitive Aspects of Web Site Navigation. Journal of the Academy of Marketing Science, Vol. 30, No. 4, pp. 397-410. Okazaki, S., 2004. Do Multinationals Standardise or Localise? The Cross-Cultural Dimensionality of Product-Based Web Sites. Internet Research, Vol. 14, No. 1, pp. 81-94. Okazaki, S. and Rivas, J., 2002. A Content-Analysis of Multinationals’ Web Communication Strategies. Cross-Cultural Research Framework and Pre-testing. Internet Research, Vol. 12, No. 5, pp. 380-390. 254 IADIS International Conference e-Commerce 2009 Rigby, D. and Vishwanath, V. 2006. Localization. The Revolution in Consumer Markets. Harvard Business Review, Vol. 84, No. 4, pp. 82-93. Robbins, S. and Stylianou, A., 2003. Global Corporate Web Sites. An Empirical Investigation of Content and Design. Information and Management, Vol. 40, No. 3, pp. 205-212. Schwartz, S., 1994. Beyond Individualism/Collectivism. New Cultural Dimensions of Values, in Kim, U. et al (ed.) Individualism and collectivism. Theory, Method, and Applications. Sage, Thousand Oaks, California. Singh, N. and Baack, D., 2004. Web Site Adaptation. A Cross-Cultural Comparison of U.S. and Mexican Web Sites. Journal of Computer-Mediated Communication, Vol. 9, No. 4. Singh, N. and Matsuo, H., 2004. Measuring Cultural Adaptation on the Web. A Content Analytic Study of U.S. and Japanese Web Sites. Journal of Business Research, Vol. 57, No. 8, pp. 864-872. Singh, N. et al, 2004. To Localize or to Standardize on the Web. Empirical Evidence from Italy, India, Netherlands, Spain, and Switzerland. Multinational Business Review, Vol. 12, No. 1, pp. 69-87. Singh, N. et al, 2005. Adaptation of Cultural Content. Evidence from B2C E-commerce Firms. European Journal of Marketing, Vol. 39, No. 1/2, pp.71-86. Singh, N. et al, 2006. Understanding International Web Site Usage. A Cross-National Study of German, Brazilian, and Taiwanese Online Consumers. International Marketing Review, Vol. 23, No. 1, pp. 83-97. Steenkamp, J.-P. and Geyskens, I., 2006. How Country Characteristics Affect the Perceived Value of Web Sites. Journal of Marketing, Vol. 70, No. July, pp. 136-150. Taylor, C., 2005. Moving International Advertising Research Forward. Journal of Advertising, Vol. 34, No. 1, pp. 7-16. Theodosiou, M. and Leonidou, L., 2003. Standardization versus Adaptation of International Marketing Strategy. An Integrative Assessment of the Empirical Research. International Business Review, Vol. 2, No. 2, pp. 141-171. Usunier, J.-C. and Lee, J., 2005. Marketing Across Cultures. Pearson Education, Harlow, England Warden, C. et al, 2002. How Worldwide Is Marketing Communication on the World Wide Web? Journal of Advertising Research, Vol. 42, No. 5, pp. 72-84. 255 ISBN: 978-972-8924-89-8 © 2009 IADIS 256 AUTHOR INDEX Abdolvand, N. ....................................................3 Afaghzadeh, S. ...................................................3 Albadvi, A. .........................................................3 Aleid, F.............................................................11 Alimazighi, Z. ................................................229 Alturas, B. ......................................................183 Andrade, P......................................................183 Bigdeli, O. ..........................................................3 Blanco, J. ..........................................................97 Bouchbout , K.................................................229 Breitsohl, J......................................................205 Brengman, M..................................................251 Briz, J. ..............................................................19 Chung, L...........................................................27 Chung, Y. .......................................................139 Cortimiglia, M. ...............................................131 Couto, J................................................... 188, 224 Cruz, R. ............................................................61 Cunha, R...........................................................61 Darooei, A. .....................................................199 de Felipe, I........................................................19 de Troyer, O. ..................................................251 do Prado, A.......................................................97 Duarte, P. .......................................................113 Eck, P. ..............................................................35 Encheva, S......................................................242 Escofet, E. ........................................................27 Esteban-Millat, I.............................................247 Fairweather, B. .................................................11 Fernández, M....................................................19 Gaedke, Y.......................................................219 Garrido, J..........................................................27 Grechenig, T.....................................................77 Higgins, S. ......................................................123 Holt, D..............................................................53 Hwang, C........................................................105 Kajan, E. ..........................................................69 Kantardzic, M.................................................123 Khammash, M. ...............................................205 Khayyambashi, M. .........................................199 Khiaonarong.T................................................215 King, D. ..........................................................123 Labidi, S. ..........................................................61 Lehrner, T.......................................................193 Leitner, P. .........................................................77 Lin, F.............................................................. 105 Löffler, C........................................................ 210 López-Herrera, A. ................................. 237, 239 Lozitskiy, O.................................................... 123 Martins, F....................................................... 163 Meseguer, A. .................................................. 247 Miguel, R. ...................................................... 113 Muñoz-Leiva, F...................................... 237, 239 Neto, P.............................................................. 61 Pohn, B........................................................... 193 Ponisio, M. ....................................................... 35 Rangone, A. ................................................... 131 Renga, F. ........................................................ 131 Ribeiro, L. ...................................................... 113 Riemens, L. ...................................................... 35 Robra-Bissantz, S........................................... 171 Rodríguez, M. .................................................. 27 Rodríguez-Ardura, I. ...................................... 247 Rogerson, S. ..................................................... 11 Sánchez-Fernández, J............................ 237, 239 Schranz, M. .................................................... 193 Shepitsen, A. .................................................. 147 Stoimenov, L.................................................... 69 Swatman, P. .................................................... 53 Takahashi, M.................................................... 45 Tan , W............................................. 89, 139, 155 Tan, Y. ..................................................... 89, 155 Terano, T.......................................................... 45 Tiago, F. ................................................. 188, 224 Tiago, M................................................. 188, 224 Tomuro, N...................................................... 147 Torres, A. ....................................................... 163 Tsuda, K........................................................... 45 Viedma-del-Jesus, M.............................. 237, 239 Vyncke, F....................................................... 251 Walgampaya, C.............................................. 123 Weinmann, M................................................. 219 Wenerstrom, B. ............................................. 123 Wilkins, L. ....................................................... 53

Log In

E-Procurement Adoption by Suppliers: A Research Proposal

E-Procurement Adoption by Suppliers: A Research Proposal

Related Papers

RELATED PAPERS