A Multi-criteria Recommendation Method for Interval Scaled Ratings

An increasing number of websites, such as Trip Advisor, reflecting the multilateral nature of their products, provide visitors with the possibility to evaluate each item on more than one criterion. The rating scale used usually is represented by a lexical or symbolic scale, such as the five-star rating system. Such scales can be regarded as interval because the symbolic or lexical semantics convey information about the strength of user preferences in addition to their order. In this paper we present I-Rec, a new multi-criteria, model-based recommender system tailored to interval scale ratings, which provides improved results over similar state of the art multi-criteria methods.