Relevance Feedback Models for Recommendation

Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing (EMNLP 2006), pages 449–456, Sydney, July 2006. c⃝2006 Association for Computational Linguistics Relevance Feedback Models for Recommendation Masao Utiyama National Institute of Information and Communications Technology 3-5 Hikari-dai, Soraku-gun, Kyoto 619-0289 Japan mutiyama@nict.go.jp Mikio Yamamoto University of Tsukuba, 1-1-1 Tennodai, Tsukuba, 305-8573 Japan myama@cs.tsukuba.ac.jp Abstract We extended language modeling ap- proaches in information retrieval (IR) to combine collaborative filtering (CF) and content-based filtering (CBF). Our ap- proach is based on the analogy between IR and CF, especially between CF and rel- evance feedback (RF). Both CF and RF exploit users’ preference/relevance judg- ments to recommend items. We first in- troduce a multinomial model that com- bines CF and CBF in a language modeling framework. We then generalize the model to another multinomial model that approx- imates the Polya distribution. This gener- alized model outperforms the multinomial model by 3.4% for CBF and 17.4% for CF in recommending English Wikipedia articles. The performance of the gener- alized model for three different datasets was comparable to that of a state-of-the- art item-based CF method. 1 Introduction Recommender systems (Resnick and Varian, 1997) help users select particular items (e.g, movies, books, music, and TV programs) that match their taste from a large number of choices by providing recommendations. The systems ei- ther recommend a set of N items that will be of interest to users (top-N recommendation problem) or predict the degree of users’ preference for items (prediction problem). For those systems to work, they first have to aggregate users’ evaluations of items explicitly or implicitly. Users may explicitly evaluate certain movies as rating five stars to express their prefer- ence. These evaluations are used by the systems as explicit ratings (votes) of items or the systems infer the evaluations of items from the behavior of users and use these inferred evaluations as implicit ratings. For example, systems can infer that users may like certain items if the systems learn which books they buy, which articles they read, or which TV programs they watch. Collaborative filtering (CF) (Resnick et al., 1994; Breese et al., 1998) and content-based (or adaptive) filtering (CBF) (Allan, 1996; Schapire et al., 1998) are two of the most popular types of algorithms used in recommender systems. A CF system makes recommendations to current (active) users by exploiting their ratings in the database. User-based CF (Resnick et al., 1994; Herlocker et al., 1999) and item-based CF (Sarwar et al., 2001; Karypis, 2001), among other CF algo- rithms, have been studied extensively. User-based CF first identifies a set of users (neighbors) that are similar to the active user in terms of their rat- ing patterns in the database. It then uses the neigh- bors’ rating patterns to produce recommendations for the active user. On the other hand, item-based CF calculates the similarity between items before- hand and then recommends items that are similar to those preferred by the active user. The perfor- mance of item-based CF has been shown …

Content truncated to 500 words