POI Recommendation based on Social Influence in Location-based Social Networks

The recommendation of Points of Interest (POIs) has been an important feature on location-based social networks, which recommends unvisited POIs to users based on their check-in history and other information from the network. In the development of POI recommender systems, many factors can influence users' decision-making process, including social influence and geographical location. Numerous approaches have been proposed in the literature to break this problem. These methods usually incorporate integrated approaches and treat the whole space of POIs as one. Nevertheless, through precisely examining two real-world check-in data, we find that in LBSNs, the friends' check-in information of POIs influences users' decision to check-in. Thus, we include social influence in recommendations and capture user preferences for POIs with social influence from friends. Moreover, we argue that social influence has a greater impact on POI recommendations than geographical influence does. Accordingly, our research shows that, contrary to previous research, users are not always interested in locations nearby, and some locations are very attractive, so users visit this POIs regardless of physical distance. In this paper, we use social influence-based friends to improve POI recommendation accuracy, and we have made the recommendation based on top-k recommendations. The effectiveness and advantage of our approach have been demonstrated in expansive trials on real-world datasets.

کلیدواژه ها:

Location-Based Social Networks ، Point-of-Interest Recommender Systems ، Social Influence ، Implicit Feedback