Time Decay Friend Recommender System for Social Network

Abstract

With the rapid development of the Internet, the recommendation system is to solve the problem of information overload and the emergence of a personalized information filtering technology. Collaborative Filtering is the most successful and widely technologies to date. But user's interests are not static, it will vary over time, and in which the social network as well. This paper selects a single forgetting function algorithm to improve research, but different users forgetting rules are not the same, it is the inadequacy of this algorithm. At the same time, the selection of experimental data object is not targeted, it cannot fully considering the characteristic of social platform for all users. The experimental results show that the proposed algorithm can efficiently improve recommendation quality.