Periodic and Successive Point-of-interest Recommendation under Dual Social Group Influences with Matrix Factorization

Abstract

Successive POI recommendation is a newly emerged research direction in recent years, which tries to recommend new POIs that the user has not visited before under the premise of knowing the current POIs of users. Previous work on successive POI recommendation only made limited improvements from using the geographical context and social information and so on. However, these approaches are mostly roughly regarded social influences as neighbor relationships and have not deeply studied social influences. Hence, in this paper, we propose a unified POI recommendation model, which is called the periodic and successive point-of-interest recommendation based on the dual social influence. We subdivide the social influence into two social groups, namely, the direct social group with link relationship and the indirect social group with the common checked-in POIs. A stochastic gradient descent based algorithm is adopted to learn the matrix factorization and the experimental study on two real-world datasets demonstrate the effectiveness of the proposed approach over the existing state-of-the-art ones.