Video on demand recommender system for internet protocol television service based on explicit information fusion

Abstract

Internet protocol television (IPTV) provides video on demand (VOD), internet service, and real-time broadcasting to users as a service that combines broadcasting and communication technology. Among various services, the sales of VOD are profitable because VODs offer relatively strong direct revenue models in IPTV services. However, the development of a VOD recommender system for IPTV service is highly challenging owing to the lack of explicit preference information of users in an IPTV environment. Previous studies for IPTV VOD recommender systems have attempted to solve the data sparsity problem through implicit preference information; however, it is better to utilize explicit preference information to improve the performance of system. Recently, IPTV service providers have provided their own over-the-top (OTT) services such that explicit preference information of users for items can be combined. Therefore, we proposed a novel information fusion method for an IPTV VOD recommender system that integrates the explicit information of both IPTV and OTT services. In addition, we utilized the probabilistic matrix factorization, that guarantees high performance in most recommender systems, as a recommender algorithm in this study. Finally, we conducted comparative evaluations based on various metrics and validated that the information fusion of IPTV and OTT services contribute to the IPTV VOD recommender system.