Research and implementation of POI recommendation system integrating temporal feature

Point-of-Interest is an interesting research topic of personalized recommendation. In this paper, a new position-based recommendation framework is proposing by using the time attribute. We establish a user matrix with user check-in data sets that contain time and location information. The user matrix is divided into the sub-matrices according to the set period and then every sub-matrix is decomposed into the user checking preference matrix and position feature matrix by using the non-negative matrix factorization. The user's attendance preferences at each time are integrated to get the user's final check-in preference for the candidate location to make recommendations more consistent with the needs of users.