Recommender Systems: An Introduction

Half-title......Page 3
Title......Page 5
Copyright......Page 6
Contents......Page 7
Foreword......Page 11
Preface......Page 15
1 Introduction......Page 19
1.1.1 Collaborative recommendation......Page 20
1.1.2 Content-based recommendation......Page 21
1.1.3 Knowledge-based recommendation......Page 22
1.1.4 Hybrid approaches......Page 24
1.1.6 Evaluating recommender systems......Page 25
1.2 Part II: Recent developments......Page 26
PART I: Introduction to basic concepts......Page 29
2.1 User-based nearest neighbor recommendation......Page 31
2.1.1 First example......Page 32
2.1.2 Better similarity and weighting metrics......Page 34
2.1.3 Neighborhood selection......Page 35
2.2 Item-based nearest neighbor recommendation......Page 36
2.2.1 The cosine similarity measure......Page 37
2.2.2 Preprocessing data for item-based filtering......Page 38
2.3.1 Implicit and explicit ratings......Page 40
2.3.2 Data sparsity and the cold-start problem......Page 41
2.4.1 Matrix factorization/latent factor models......Page 44
2.4.2 Association rule mining......Page 49
2.4.3 Probabilistic recommendation approaches......Page 53
2.5 Recent practical approaches and systems......Page 58
2.5.1 Slope One predictors......Page 59
2.5.2 The Google News personalization engine......Page 62
2.6 Discussion and summary......Page 65
2.7 Bibliographical notes......Page 67
3 Content-based recommendation......Page 69
3.1 Content representation and content similarity......Page 70
3.1.1 The vector space model and TF-IDF......Page 72
3.1.2 Improving the vector space model/limitations......Page 74
3.2.1 Nearest neighbors......Page 76
3.2.2 Relevance feedback – Rocchio’s method......Page 78
3.3.1 Probabilistic methods......Page 81
3.3.2 Other linear classifiers and machine learning......Page 86
3.3.3 Explicit decision models......Page 88
3.3.4 On feature selection......Page 90
3.4.1 Comparative evaluation......Page 92
3.4.2 Limitations......Page 93
3.5 Summary......Page 95
3.6 Bibliographical notes......Page 97
4.1 Introduction......Page 99
4.2 Knowledge representation and reasoning......Page 100
4.2.1 Constraints......Page 101
4.2.2 Cases and similarities......Page 104
4.3 Interacting with constraint-based recommenders......Page 105
4.3.1 Defaults......Page 106
4.3.2 Dealing with unsatisfiable requirements and empty result sets......Page 109
4.3.3 Proposing repairs for unsatisfiable requirements......Page 114
4.3.4 Ranking the items/utility-based recommendation......Page 115
4.4 Interacting with case-based recommenders......Page 119
4.4.1 Critiquing......Page 120
4.4.2 Compound critiquing......Page 122
4.4.3 Dynamic critiquing......Page 123
4.4.4 Advanced item recommendation......Page 128
4.4.5 Critique diversity......Page 129
4.5.1 The VITA constraint-based recommender......Page 131
4.5.2 The Entree case-based recommender......Page 137
4.6 Bibliographical notes......Page 140
5 Hybrid recommendation approaches......Page 142
5.1 Opportunities for hybridization......Page 143
5.1.1 Recommendation paradigms......Page 144
5.1.2 Hybridization designs......Page 146
5.2 Monolithic hybridization design......Page 147
5.2.1 Feature combination hybrids......Page 148
5.2.2 Feature augmentation hybrids......Page 150
5.3.1 Mixed hybrids......Page 152
5.3.2 Weighted hybrids......Page 153
5.3.3 Switching hybrids......Page 155
5.4.1 Cascade hybrids......Page 156
5.4.2 Meta-level hybrids......Page 157
5.5 Discussion and summary......Page 159
5.6 Bibliographical notes......Page 160
6.1 Introduction......Page 161
6.2 Explanations in constraint-based recommenders......Page 165
6.2.1 Example......Page 166
6.2.2 Generating explanations by abduction......Page 169
6.2.3 Analysis and outline of well-founded explanations......Page 171
6.2.4 Well-founded explanations......Page 173
6.3 Explanations in case-based recommenders......Page 175
6.4 Explanations in collaborative filtering recommenders......Page 179
6.5 Summary......Page 183
7.1 Introduction......Page 184
7.2 General properties of evaluation research......Page 185
7.2.1 General remarks......Page 186
7.2.2 Subjects of evaluation design......Page 187
7.2.3 Research methods......Page 190
7.3 Popular evaluation designs......Page 193
7.4.1 Methodology......Page 195
7.4.2 Metrics......Page 197
7.5 Alternate evaluation designs......Page 202
7.5.1 Experimental research designs......Page 203
7.5.3 Nonexperimental research designs......Page 204
7.6 Summary......Page 205
7.7 Bibliographical notes......Page 206
8 Case study: Personalized game recommendations on the mobile Internet......Page 207
8.1 Application and personalization overview......Page 209
8.2 Algorithms and ratings......Page 211
8.3.1 Measurement 1: “My Recommendations”......Page 212
8.3.2 Measurement 2: Post-sales recommendations......Page 215
8.3.3 Measurement 3: Start page recommendations......Page 218
8.3.4 Measurement 4: Overall effect on demo downloads......Page 220
8.3.5 Measurement 5: Overall effects......Page 221
8.4 Summary and conclusions......Page 224
PART II: Recent developments......Page 227
9 Attacks on collaborative recommender systems......Page 229
9.1 A first example......Page 230
9.2 Attack dimensions......Page 231
9.3 Attack types......Page 232
9.3.2 The average attack......Page 233
9.3.4 The segment attack......Page 234
9.3.6 Clickstream attacks and implicit feedback......Page 235
9.4.1 Push attacks......Page 237
9.5 Countermeasures......Page 239
9.6.1 Centralized methods: Data perturbation......Page 243
9.6.2 Distributed collaborative filtering......Page 245
9.7 Discussion......Page 250
10.1 Introduction......Page 252
10.2 Context effects......Page 254
10.3 Primacy/recency effects......Page 258
10.4 Further effects......Page 261
10.5 Personality and social psychology......Page 263
10.6 Bibliographical notes......Page 270
11 Recommender systems and the next-generation web......Page 271
11.1 Trust-aware recommender systems......Page 272
11.1.1 Exploiting explicit trust networks......Page 273
11.1.2 Trust metrics and effectiveness......Page 275
11.1.3 Related approaches and recent developments......Page 278
11.2 Folksonomies and more......Page 280
11.2.1.1 Folksonomies and content-based methods......Page 281
11.2.1.2 Folksonomies and collaborative filtering......Page 287
11.2.2 Recommending tags......Page 291
11.2.3 Recommending content in participatory media......Page 294
11.3.1 Augmentation of filtering by taxonomies......Page 297
11.3.2 Augmentation of filtering by attributes......Page 302
11.4 Extracting semantics from the web......Page 303
11.5 Summary......Page 306
12.1 Introduction......Page 307
12.2 Context-aware recommendation......Page 309
12.3 Application domains......Page 312
12.4 Summary......Page 315
13.1 Summary......Page 317
13.2 Outlook......Page 318
Bibliography......Page 323
Index......Page 351