Recommender Systems

Lecturer at the GERiiCO laboratory at University Lille 3, Gerald Kembellec specializes in information science and communication.Professor of Documentary Engineering Chair of CNAM, Ghislaine Chartron is director of the National Institute of Science and Technical documentation.Professor at the University Paris 8, Imad Saleh is the Paragraph laboratory director and director of the graduate school Cognition Language Interaction.

• PREFACE xiGérald KEMBELLEC, Ghislaine CHARTRON and Imad SALEHCHAPTER 1. GENERAL INTRODUCTION TO RECOMMENDER SYSTEMS 1Ghislaine CHARTRON and Gérald KEMBELLEC1.1. Putting it into perspective 11.2. An interdisciplinary subject 21.3. The fundamentals of algorithms 41.3.1. Collaborative filtering 41.3.2. Content filtering 71.3.3. Hybrid methods 91.3.4. Conclusion on historical recommendation models 111.4. Content offers and recommender systems 111.4.1. Culture and recommender systems 111.4.2. Recommender systems and the e-commerce of content 161.4.3. The behavior of users 181.5. Current issues 191.6. Bibliography 19CHAPTER 2. UNDERSTANDING USERS’ EXPECTATIONS FOR RECOMMENDER SYSTEMS: THE CASE OF SOCIAL MEDIA 25Jean-Claude DOMENGET and Alexandre COUTANT2.1. Introduction: the omnipresence of recommender systems 252.2. The social approach to prescription 272.2.1. The theory of the prescription and online interactions 272.2.2. Conditions for recognition of the prescription 292.2.3. The specificities of social media 302.3. Users who do not focus on the prescriptions of platforms 312.3.1. Facebook: the link, the type of activity and the context 322.3.2. Twitter: prescription between peers and explanation of prescription 382.3.3. Conditions for the recognition of a prescription: announcement and enunciation 442.4. A guide for considering recommender systems adapted to different forms of social media 452.5. Conclusion 482.6. Bibliography 49CHAPTER 3. RECOMMENDER SYSTEMS AND SOCIAL NETWORKS: WHAT ARE THE IMPLICATIONS FOR DIGITAL MARKETING? 53Maria MERCANTI-GUÉRIN3.1. Social recommendations: an ancient practice revived by the digital age 543.1.1. Recommendations: a difficult management for brands 553.1.2. Internet recommendations: social presence and personalized recommendations 553.2. Social recommendations: how are they used for e-commerce? 583.2.1. Efficiency of recommender systems with regard to the performance of e-commerce websites 583.2.2. Recommender systems used by social networks: from e-commerce to social commerce 593.3. Conclusion 663.4. Bibliography 68CHAPTER 4. RECOMMENDER SYSTEMS AND DIVERSITY: TAKING ADVANTAGE OF THE LONG TAIL AND THE DIVERSITY OF RECOMMENDATION LISTS 71Muriel FOULONNEAU, Valentin GROUÈS, Yannick NAUDET and Max CHEVALIER4.1. The stakes associated with diversity within recommender systems 724.1.1. Individual diversity or the individual perception of diversity 734.1.2. The stakes and impacts of aggregate diversity 744.2. Recommendation algorithms and diversity: trends, evaluation and optimization 774.2.1. The tendency for recommendation algorithms to focus on the head 784.2.2. The evaluation of diversity in recommender systems 804.2.3. Recommendation algorithms which favor individual diversity 814.2.4. Recommendation algorithms which favor aggregate diversity 814.2.5. The shift toward user-centered diversity approaches 824.3. Conclusion and new directions 854.4. Bibliography 87CHAPTER 5. ISONTRE: INTELLIGENT TRANSFORMER OF SOCIAL NETWORKS INTO A RECOMMENDATION ENGINE ENVIRONMENT 93Rana CHAMSI ABU QUBA, Salima HASSAS, Usama FAYYAD, Hammam CHAMSI and Christine GERTOSIO5.1. Summary 935.2. Introduction 945.3. Latest developments, definition and history 975.3.1. Collaborative filtering techniques 975.3.2. General use social networks: what do they contain? 975.3.3. Social recommendation 995.3.4. The recommendation of concepts 1005.4. iSoNTRE 1015.4.1. iSoNTRE: transformer of social networks 1025.4.2. iSoNTRE: the core of recommendation 1075.5. Experiments 1105.5.1. The preparation of data 1105.5.2. Testing methodology 1105.5.3. The creation of avatars 1115.5.4. Results 1125.5.5. Discussion 1135.6. Conclusion 1145.7. Bibliography 115CHAPTER 6. A TWO-LEVEL RECOMMENDATION APPROACH FOR DOCUMENT SEARCH 119Manel HMIMIDA and Rushed KANAWATI6.1. Introduction 1196.2. Tag recommendation: a brief state of the art 1206.3. The hypertagging system 1226.3.1. Metadata 1226.3.2. Architecture 1236.4. Recommendation approach 1246.4.1. Presentation 1246.4.2. Recommendation algorithm 1266.5. Evaluation 1276.5.1. Generation of facets 1276.5.2. Generation of association rules 1296.5.3. Evaluation of recommendation rules 1306.6. Conclusion 1316.7. Bibliography 132CHAPTER 7. COMBINING CONFIGURATION AND RECOMMENDATION TO ENABLE AN INTERACTIVE GUIDANCE OF PRODUCT LINE CONFIGURATION 135Raouia TRIKI , Raúl MAZO and Camille SALINESI7.1. Introduction 1357.2. Context 1377.2.1. Configuration 1377.2.2. Recommendation 1397.2.3. Obstacles and challenges of interactive PL configuration 1417.3. Overview of the proposed approach 1427.4. Preliminary evaluation 1487.5. Discussion and related work 1487.5.1. Recommendation techniques 1487.6. Conclusion and future work 1517.7. Bibliography 151CHAPTER 8. SEMIO-COGNITIVE SPACES: THE FRONTIER OF RECOMMENDER SYSTEMS 157Hakim HACHOUR, Samuel SZONIECKY and Safia ABOUAD8.1. Introduction 1578.2. Latest developments: finalized activities, recommender systems and the relevance of information 1598.2.1. Cognitive dynamics of finalized activities 1598.2.2. The foundations of recommender systems 1618.2.3. What information relevance? 1668.3. Observable interests for decision theory: a combination of content-based, collaboration based and knowledge-based recommendations 1698.3.1. Methodology: meta-analysis and modeling of the process 1698.3.2. Analysis and modeling of a macro-process for responding to a call for R&D projects 1718.3.3. Analysis and model of a socio-organizational tool for the management of customer complaints 1738.4. Discussion and conclusions 1778.4.1. Discussion: the performance of the filtering methods and semio-cognitive criteria for relevance 1778.5. Conclusions: recommender systems linked to finalized activities 1818.5.1. The localization of activities and geographical information systems: a new kind of data 1828.5.2. Transparency of the use of personal data, data protection and ownership 1838.6. Acknowledgments 1858.7. Bibliography 185CHAPTER 9. THE FRENCH-SPEAKING LITERARY PRESCRIPTION MARKET IN NETWORKS 191Louis WIART9.1. Introduction 1919.2. The economy of prescription 1939.2.1. The notion of prescription 1939.2.2. From the advisors market to the prescription market 1949.3. Methodology 1969.4. The competitive structure of the market of online social networks of readers 1979.4.1. Pure player networks and the audience strategy 1999.4.2. Amateur networks and the survival strategy 2019.4.3. Backed networks and the hybridization strategy 2029.5. The organization of prescription 2049.5.1. Social prescription 2059.5.2. Editorial prescription 2069.5.3. Algorithmic prescription 2079.6. Conclusion: what legitimacy for literary prescription? 2089.7. Appendix: list of interviews undertaken 2109.8. Bibliography 210CHAPTER 10. PRESENTATION OF OFFERED SERVICES: BABELIO, A RECOMMENDATION ENGINE DEDICATED TO BOOKS 213Vassil STEFANOV, Guillaume TEISSEIRE and Pierre FRÉMAUX10.1. Introduction 21310.2. The problem of qualitative pertinence 21610.3. The problem of quantitative pertinence 21710.4. Balancing recall and precision 21710.5. The issue of sparse data 21810.6. Performance and scalability 21810.7. A few issues specific to books 219CHAPTER 11. PRESENTATION OF THE OFFER OF SERVICES: NOMAO, RECOMMENDER SYSTEMS AND INFORMATION SEARCH 221Estelle DELPECH, Laurent CANDILLIER and Étienne CHAI11.1. Introduction: the actors of Internet recommendation 22111.2. Approaches to recommendation 22211.3. Nomao: a local outlets search and recommendation engine 22311.3.1. Popularity score 22311.3.2. Affinity score 22411.3.3. Social recommendation 22511.4. Prospects: the move toward interactive recommender systems 22511.5. Appendix 226LIST OF AUTHORS 227INDEX 231