Point-of-Interest Recommendation Based on Spatial Clustering in LBSN

In location-based social networks, many studies have been put into forward to improve point-of-interest (POI) recommendation, according to the users' historical check-ins and context aware information. But the spatial distribution feature of the users' check-in has not been well studied. We propose a POI recommendation algorithm based on spatial clustering, through studying whether the users' check-in behaviors have "active area" differences. Firstly, a spatial clustering algorithm is designed according to the administrative area information of the POI in the location-based social networks, which is combined with users' check-in distribution characteristics. Secondly, the users' ratings information is normalized. Finally, POI recommendation is obtained according to integration each user's check-in spatial clustering subset distribution and user's ratings to social relationships. The experimental results show that the significant improvement and the effectiveness of the method in the precision, recall rate and time performance.