A Multi-factor Recommendation Algorithm for POI Recommendation.

Point-of-Interest (POI) recommendation is an important service in Location-Based Social Networks (LBSNs). There are several approaches, such as collaborating filtering or content-based filtering, to solving the problem, but the quality of recommendation is low because of lack of personalized influencing factors for each user. In LBSNs, users’ history check-in data contain rich information, such as the geographic and textual information of the POIs, the time user visiting POIs, the friend relationship between users, etc. Recently, these factors are exploited to further improve the quality of recommendation. The major challenges are which factors to use and how to use them. In this paper, a multi-factor recommendation algorithm (MFRA) is proposed to address these challenges, which initially exploits the locality of user activities to create a candidate set of POIs, and for each candidate POI, the influence of friend relationship and the category and popularity of POI on each user are considered to improve the quality of recommendation. Experiments on the check-in datasets of Foursquare demonstrate a better precision and recall rate of the proposed algorithm.