Abstract: In the location-based social network,collaborative filtering is the most widely used recommended technology,but it has some drawbacks,such as data sparsity and cold start.In light of this,this paper presented a point of interest(POI) recommendation algorithm combining user’s interest and geographic factors.In this method,firstly,the adaptive kernel density distribution,naive Bayesian algorithm and the popularity of POIs are combined to mine user’s geographi-cal preferences,and some candidate recommended POIs are screened out according to the geographical preference model.Then,in order to overcome the problems of data sparsity and cold start in collaborative filtering algorithm,a user pre-ference model is constructed to carry out POI recommendation based on the similarities of user checked-in,category information and user trust.Finally,the Yelp data set was used to conduct the experimental analysis.The results show that the proposed POI recommendation model based on user’s interest and geographical factor obtains good recommendation effect.

Key words: Category information, Collaborative filtering, Geographical preferences, POI recommendation, Trust relationship