A General Architecture for an Emotion-aware Content-based Recommender System
Fedelucio Narducci, Marco de Gemmis, Pasquale Lops · 2015
Emotions play a crucial role in the decision making process. Frequently, choices are strongly influenced by the mood of the moment, and the same person could take different decisions at different time on the same topic. Recommender systems, that are definitively recognized as tools for supporting the decision making process, demonstrated to be more accurate exploiting emotive labels in several work. For this reason a large number of researchers are focusing their attention on the analysis of the emotions by exploiting data that users daily disseminate on the Web (e.g.: Social Networks, Blogs, Forums, etc.). In this paper we propose a general architecture for implementing an emotion-aware content-based recommender system. Furthermore, we developed a web service that researchers can freely exploit for their own implementations. We carried out a user study on the domain of music recommendation, particularly influenced by the user emotion, and results are very promising.