The Rapid Development of the mobile based social networks is to improve the geo-graphical rating procedure, which makes enormous volume of the raw data. In normal mobile user compared to recent trends of the geo-position enabled technology will produce the volume of data is higher than the normal mobile user. The social networks involving geographical information as location-based social networks, this information bring out the new challenge in recommended system to solve the data sparsity problem of data set and cold start problem, in this paper we make the full use of the deeply exploring the user and check-in user for various categories first, user to user geographical connection distance, then user to item geographical distance and one user similarities. Check-in behaviors of users will be deeply explored by considering the above factor their multi-activity centers and the attribute of POIs.