Folksonomy-based fuzzy user profiling for improved recommendations

• Fuzzy logic and linguistic quantifier-driven determination of representative tags for genre is proposed.
• A novel approach to deriving the extent of genre presence in movies by leveraging on folksonomy data is presented.
• The reclusive method is revised to automatically derive user preferred content based features.
• Different ways of utilizing the fuzzy user and item profiles are explored for performing recommendations.

Genre is a major factor influencing user decisions to peruse an item in domains such as movies, books etc. Recommender systems, generally have, at their disposal, information regarding genres/categories that a movie/book belongs to. However, the degree of membership of the objects in these categories is typically unavailable. Such information, if available, would provide a better description of items and consequently lead to quality recommendations. In this paper, we propose an approach to infer the degree of genre presence in a movie by examining the various tags conferred on them by various users. Tags are user-defined metadata for items and embed abundant information about various facets of user likes, their opinion on the quality and the type of object tagged. Leveraging on tags to guide the genre degree determination exploits crowd sourcing to enrich item content description. Fuzzy logic naturally models human logic allowing for the nuanced representation of features of objects and thus is utilized to derive such gradual representation as well as for modeling user profiles. To the best of our knowledge ours is one of the first approaches to utilize such folksonomy information to infer genre degrees subsequently used for recommendations. The proposed method has the twin advantages of utilizing enriched content information for recommendation as well as squeezing the information from the user-item-tag and user-item ratings spaces and condensing them into fuzzy user profiles. The fuzzy user and object representations are leveraged both for the design of content-based as well as collaborative recommender systems. Experimental evaluations establish the effectiveness of the proposed approaches as compared to other baselines.