Fuzzy Rule-Based Situational Music Retrieval and Recommendation

Abstract

Recommender Systems (RS) have been used in many domains to assist users' decision making by providing item recommendations and by reducing information overload. Context-Aware Recommender System (CARS) is one RS trying to generate more convincing recommendations by taking user's context into regard. Thus, in order to adapt the recommendations to the user's context, it is useful to identify the current situation of the involved user which may influence the acceptance of a recommendation. Improving user's acceptance on recommendations is a challenging task. In recommender systems, determining the right contextual situation and finding relevant items for the user are considered as two vital issues for achieving better user acceptance. Therefore, our ultimate aim is to infer user's actual situation by fusing interacted contextual circumstances surrounding music listeners through Fuzzy Logic. Then, the inferred situation is used to recommend music from an online webradio service. We conducted a case study to assess user satisfaction in music recommendation. Experimental results reveal that the inclusion of the inferred situation increases significantly the accuracy of the recommendation, while compared to the traditional contextual recommender systems.