22. 4.1 Data Collection and Preprocessing 722 22.4.2 Desired Properties, Semantics and Interpretation 724 22. 4.3 Complexity and the Understanding of Function behavior 22. 4.4 Weight and Parameter Determination 22.5 Sophisticated Aggregation Procedures in Recommender Systems: Tailoring for Specific Applications 726 22.6 Conclusions 731 22. 7 Further readins References 733 23 Active Learning in Recommender Systems Neil Rubens, Dain Kaplan, and Masashi Sugiyama 3.1 Introduction 735 3. 1. 1 Objectives of Active Learning in Recommender Systems 737 3. 1.2 An Illustrative Example 23. 1.3 Types of Active Learning 739 23.2 Properties of Data Points 23.2.1 Other Considerations 741 23.3 Active Learning in Recommender Systems 742 23.3.1 Method Summary Matrix 23.4 Active Learning Formulation 742 23.5 Uncertainty-based Active Learnin 746 23.5.1 Output Uncertainty 23.5.2 Decision Boundary Uncertainty 23.5.3 Model Uncertainty 749 23.6 Error-based Active Learning 3.6.1 Instance-based Methods 23 6.2 Model-based 23.7 Ensemble-based Active Learning 23.7.1 Models-based 756 23. 8 Conversation-based Active Learnin 23.8.1 Case-based Critique 23.8.2 Diversity-based 23.8.3 Query Editing-based 762 23.9 Computational Considerations 762 3.10 Discussion References 764 24 Multi-Criteria Recommender Systems 769 Gediminas Adomavicius, Nikos Manouselis and YoungOk Kwon 769 24.2 Recommendation as a multi-Criteria decision Making Problem 24.2. 1 Object of Decision 4.2.2 Family of Criteria 773
xx Contents 22.4.1 Data Collection and Preprocessing . . . . . . . . . . . . . . . . . . . 722 22.4.2 Desired Properties, Semantics and Interpretation . . . . . . . 724 22.4.3 Complexity and the Understanding of Function Behavior 725 22.4.4 Weight and Parameter Determination . . . . . . . . . . . . . . . . . 726 22.5 Sophisticated Aggregation Procedures in Recommender Systems: Tailoring for Specific Applications . . . . . . . . . . . . . . . . . . . 726 22.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 731 22.7 Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 732 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733 23 Active Learning in Recommender Systems . . . . . . . . . . . . . . . . . . . . . . 735 Neil Rubens, Dain Kaplan, and Masashi Sugiyama 23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 735 23.1.1 Objectives of Active Learning in Recommender Systems 737 23.1.2 An Illustrative Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . 738 23.1.3 Types of Active Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . 739 23.2 Properties of Data Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 740 23.2.1 Other Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 741 23.3 Active Learning in Recommender Systems . . . . . . . . . . . . . . . . . . . . 742 23.3.1 Method Summary Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . 742 23.4 Active Learning Formulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 742 23.5 Uncertainty-based Active Learning . . . . . . . . . . . . . . . . . . . . . . . . . . 746 23.5.1 Output Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 746 23.5.2 Decision Boundary Uncertainty . . . . . . . . . . . . . . . . . . . . . . 748 23.5.3 Model Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749 23.6 Error-based Active Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751 23.6.1 Instance-based Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 752 23.6.2 Model-based . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 754 23.7 Ensemble-based Active Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756 23.7.1 Models-based . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 756 23.7.2 Candidates-based . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757 23.8 Conversation-based Active Learning . . . . . . . . . . . . . . . . . . . . . . . . . 760 23.8.1 Case-based Critique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761 23.8.2 Diversity-based . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761 23.8.3 Query Editing-based . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 762 23.9 Computational Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 762 23.10 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 763 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 764 24 Multi-Criteria Recommender Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 769 Gediminas Adomavicius, Nikos Manouselis and YoungOk Kwon 24.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769 24.2 Recommendation as a Multi-Criteria Decision Making Problem 771 24.2.1 Object of Decision. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 772 24.2.2 Family of Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 773 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Contents 2 3 Global Preference model 24.2.4 Decision Support Process 775 24.3 MCDM Framework for Recommender Systems Lessons learned 776 24 4 Multi-Criteria Rating recommendation 24.4. 1 Traditional single-rating recommendation problem 781 4.4.2 Extending traditional recommender systems to include 24.5 Survey of Algorithms for Multi-Criteria Rating Recommenders.. 783 24.5.1 Engaging Multi-Criteria Ratings during Prediction... 784 24.5.2 Engaging multi-Criteria Ratings during Recommendation 791 24.6 Discussion and Future Work 795 24.7 Conclusions 25 Robust Collaborative Recommendation Robin Burke, Michael P O Mahony and Neil J. Hurley 25.1 Introduc 805 25.2 Defining the Problem 807 25.2.1 An Example Attack 25.3 Characterising Attacks 25.3.1 Basic Attacks 810 25.3.2 Low-knowledge attacks 25.33 Nuke Attack models 25.3.4 Informed Attack Models 25.4 Measuring Robustness 814 25.42 Push Attacks 816 25.43 Nuke Attacks 818 5.4.4 Informed Attacks 25.45 Attack 25.5 Attack Detection 5.5.1 Evaluation metrics 821 25.5. 2 Single profile detection 5.5.3 Group Profile Detection 25.5.4 Detection findings 25.6 Robust Algorithms 5.6.1 Model-based Recomendation 25.6.2 Robust Matrix Factorisation(RMF) 25.6.3 Other Robust Recommendation Algorithms 25.64 The Influence Limiter and Trust-based Recommendation 831 onc References 833 Index 837
Contents xxi 24.2.4 Decision Support Process . . . . . . . . . . . . . . . . . . . . . . . . . . . 775 24.3 MCDM Framework for Recommender Systems: Lessons Learned 776 24.4 Multi-Criteria Rating Recommendation . . . . . . . . . . . . . . . . . . . . . . . 780 24.4.1 Traditional single-rating recommendation problem . . . . . . 781 24.4.2 Extending traditional recommender systems to include multi-criteria ratings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 782 24.5 Survey of Algorithms for Multi-Criteria Rating Recommenders . . . 783 24.5.1 Engaging Multi-Criteria Ratings during Prediction . . . . . . 784 24.5.2 Engaging Multi-Criteria Ratings during Recommendation 791 24.6 Discussion and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795 24.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 797 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 798 25 Robust Collaborative Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . 805 Robin Burke, Michael P. O’Mahony and Neil J. Hurley 25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 805 25.2 Defining the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 807 25.2.1 An Example Attack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 809 25.3 Characterising Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810 25.3.1 Basic Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 810 25.3.2 Low-knowledge attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 811 25.3.3 Nuke Attack Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 812 25.3.4 Informed Attack Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 813 25.4 Measuring Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 814 25.4.1 Evaluation Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815 25.4.2 Push Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 816 25.4.3 Nuke Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 818 25.4.4 Informed Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 819 25.4.5 Attack impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 820 25.5 Attack Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 820 25.5.1 Evaluation Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 821 25.5.2 Single Profile Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 822 25.5.3 Group Profile Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 824 25.5.4 Detection findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 827 25.6 Robust Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 828 25.6.1 Model-based Recomendation . . . . . . . . . . . . . . . . . . . . . . . . 828 25.6.2 Robust Matrix Factorisation (RMF) . . . . . . . . . . . . . . . . . . 829 25.6.3 Other Robust Recommendation Algorithms . . . . . . . . . . . . 830 25.6.4 The Influence Limiter and Trust-based Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 831 25.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 832 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 833 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 837 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24.2.3 Global Preference Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 774
List of contributors Gediminas Adomavicius Department of Information and Decision Sciences Carlson School of Management, University of Minnesota, Minneapolis, MN 55455, USA e-mail: gedas@umn. edu Xavier amatriain Telefonica Research, Via Augusta, 122, Barcelona 08021, Spain e-mail: xar@tides Riccardo bambini Fastweb, via Francesco Caracciolo 51, Milano, Italy e-mail: riccardo bambini fastweb. it Gleb beliakoy School of Information Technology, Deakin University, 221 Burwood Hwy. Burwood 3125. australia e-mail: gleb@deakin. edu.au at&T Labs-Research e-mail:rbell@research.att.com David Bonnefoy Pearltrees e-mail:davidbonnefoypearltrees.com Peter briggs CLARITY: Centre for Sensor Web Technologies, School of Computer Science Informatics, University College Dublin, Ireland, e-mail: Peter Briggs@ucd.ie Robin burk Center for Web Intelligence, School of Computer Science, Telecommunication and
List of Contributors Gediminas Adomavicius Department of Information and Decision Sciences Carlson School of Management, University of Minnesota, Minneapolis, MN 55455, USA e-mail: gedas@umn.edu Xavier Amatriain Telefonica Research, Via Augusta, 122, Barcelona 08021, Spain e-mail: xar@tid.es Riccardo Bambini Fastweb, via Francesco Caracciolo 51, Milano, Italy e-mail: riccardo.bambini@fastweb.it Gleb Beliakov School of Information Technology, Deakin University, 221 Burwood Hwy, Burwood 3125, Australia, e-mail: gleb@deakin.edu.au Robert Bell AT&T Labs – Research e-mail: rbell@research.att.com David Bonnefoy Pearltrees, e-mail: david.bonnefoy@pearltrees.com Peter Briggs CLARITY: Centre for Sensor Web Technologies, School of Computer Science & Informatics, University College Dublin, Ireland, e-mail: Peter.Briggs@ucd.ie Robin Burke Center for Web Intelligence, School of Computer Science, Telecommunication and xxiii
List of contributors Information Systems, DePaul University, Chicago, Illinois, USA Tomasa Calvo Departamento de Ciencias de la Computacion, Universidad de alcala 28871-Alcala de Henares(Madrid), Spain e-mail: tomasa calvo uah.es Li Chen Human Computer Interaction Group, School of Computer and Communication Sciences Swiss Federal Institute of Technology in Lausanne(EPFL), CH-1015, Lausanne, Switzerland e-mail: lichen @epfl.ch Martine de cock Institute of Technology, University of Washington Tacoma, 1900 Pacific Ave, Tacoma, WA, USA (on leave from Ghent University) Chris Cornelis Dept of Applied Mathematics and Computer Science, Ghent University, Krijgslaan 281(S9), 9000 Gent, Belgium e-mail: Patricia. Victor@ugent. be Maurice Coyle CLARITY: Centre for Sensor Web Technologies, School of Computer Science Informatics, University College Dublin, Ireland, -mail: Maurice. Coyle@ucd. ie Paolo cremonesi Politecnico di Milano, p zza Leonardo da vinci 32, Milano, Italy Neptuny, via Durando 10, Milano, Italy e-mail: paolo cremonesi@polimiit Christian Desrosiers epartment of Software Engineering and IT, Ecole de Technologie Superieure montreal Canada e-mail: christian desrosiers @ etsmtlca Hendrik drachsler Centre for Learning Sciences and Technologies(CELSTEC), Open Universiteit Nederland e-mail: hendrik drachsler(@ou. nl Artificial Intelligence Laboratory, School of Computer and Communication Sciences Swiss Federal Institute of Technology in Lausanne(EPFL), CH-1015, Lausanne Switzerland e-mail: boi faltings@epfl. ch
xxiv List of Contributors Information Systems, DePaul University, Chicago, Illinois, USA e-mail: rburke@cs.depaul.edu Tomasa Calvo Departamento de Ciencias de la Computacion, Universidad de Alcal ´ a´ 28871-Alcala de Henares (Madrid), Spain. ´ e-mail: tomasa.calvo@uah.es Li Chen Human Computer Interaction Group, School of Computer and Communication Sciences, Swiss Federal Institute of Technology in Lausanne (EPFL), CH-1015, Lausanne, Switzerland e-mail: li.chen@epfl.ch Martine De Cock Institute of Technology, University of Washington Tacoma, 1900 Pacific Ave, Tacoma, WA, USA (on leave from Ghent University) e-mail: mdecock@u.washington.edu Chris Cornelis Dept. of Applied Mathematics and Computer Science, Ghent University, Krijgslaan 281 (S9), 9000 Gent, Belgium e-mail: Patricia.Victor@ugent.be Maurice Coyle CLARITY: Centre for Sensor Web Technologies, School of Computer Science & Informatics, University College Dublin, Ireland, e-mail: Maurice.Coyle@ucd.ie Paolo Cremonesi Politecnico di Milano, p.zza Leonardo da Vinci 32, Milano, Italy Neptuny, via Durando 10, Milano, Italy e-mail: paolo.cremonesi@polimi.it Christian Desrosiers Department of oftware Engineering and I ,T Ecole de Technologie Superieure, ´ ´ Montreal, e-mail: christian.desrosiers@etsmtl.ca Hendrik Drachsler Centre for Learning Sciences and Technologies (CELSTEC), Open Universiteit Nederland e-mail: hendrik.drachsler@ou.nl Boi Faltings Artificial Intelligence Laboratory, School of Computer and Communication Sciences Swiss Federal Institute of Technology in Lausanne (EPFL), CH-1015, Lausanne, Switzerland S Canada e-mail: boi.faltings@epfl.ch