A new TV recommendation algorithm based on interest quantification and item clustering
Chao Cheng, Xingjun Wang, Zhiyong Li, Yuxi Lin · 2015
Recommender Systems(RSs) are software tools and techniques providing suggestions for items to be of use to a user. With the increasing development of Internet and explosion of information, recommender system has been an indispensable component in many applications. In this paper, a recommendation algorithm based on factorization model is proposed, which is applied to TV system. To quantize users' interest/preference to programs, a novel and rational notation, user interest index, is defined and helps improve recommendation effect. The vectorization of users and programs are derived from item clustering. Finally, we adopted top-K recommendation strategy, and evaluated the performance of our algorithm. According to experiment results, we found that the algorithm performs well on precision and recall rate.