ABSTRACT
Electrical Engineering, Mathematics and Computer Science, University of Twente, Netherlands; Rudjer Boskovic Institute, Zagreb, Croatia
Matko Bos˘njak
University of Porto, Porto, Portugal; Rudjer Boskovic Institute, Zagreb, Croatia
Nino Antulov-Fantulin
Rudjer Boskovic Institute, Zagreb, Croatia
Tomislav Sˇmuc
Rudjer Boskovic Institute, Zagreb, Croatia
Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 9.2 The Recommender Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
9.2.1 Recommendation Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 9.2.2 Data Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 9.2.3 Performance Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
9.3 The VideoLectures.net Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 9.4 Collaborative-based Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
9.4.1 Neighbourhood-based Recommender Systems . . . . . . . . . . . . . . . . . . . . . . . . 127 9.4.2 Factorization-based Recommender Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 9.4.3 Collaborative Recommender Workflows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 9.4.4 Iterative Online Updates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.5 Content-based Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 9.5.1 Attribute-based Content Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 9.5.2 Similarity-based Content Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
9.6 Hybrid Recommender Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 9.7 Providing RapidMiner Recommender System Workflows as Web Services
Using RapidAnalytics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 9.7.1 Simple Recommender System Web Service . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 9.7.2 Guidelines for Optimizing Workflows for Service Usage . . . . . . . . . . . . . . 139
9.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Applications
k-NN - k Nearest Neighbours ROC curve - Receiver Operating Characteristic curve AUC - Area Under the Curve Prec@k - Precision at k MAP - Mean Average Precision NDCG - Normalized Discounted Cumulative Gain
Making choices is an integral part of everyday life, especially today when users are overwhelmed with information, from the Internet and television, to shelves in local stores and bookshops. We cope with this information overload by relying on daily recommendations from family, friends, authoritative users, or users who are simply willing to offer such recommendations. This is especially important when we lack information to make a rational decision, for example, choosing a hotel for vacations in a new city, selecting a new movie to watch, or choosing which new restaurant to visit.
