Mining Semantic Location History for Collaborative POI recommendation in Online Social Networks

Location-based social networks (LBSNs) have recently attracted millions of mobile users to explore attractive locations and share their visited experiences. As the increasing use of the online social networks, people demand personalized service to recommend places of interests (POIs) based on their personal preferences. Among POIs recommendation approaches, collaborative filtering that predicts POIs of the user based on the geospatial location and users' opinions is suite for LBSNs. Despite this, it is still a challenge to infer the similarity between users because of the unique characteristics of spatial items in LBSN. In this paper, we propose an effective POI recommendation method for LBSNs based on collaborative filtering. Our method focuses on mining interest similarity of users based on their check-in activities in LBSN. Since the geospatial locations cannot capture user's interests, we perform to mine semantic features of user's check-in history based on semantic location descriptions to discover the user's interests. We finally perform recommending nearby places to a particular user by fusing opinions from similar users according to the user's current location. Experimental results with two real-world datasets collected from Foursquare show that our proposed method can achieve satisfying precision and recall of recommended places.