Research On Point-of-Interest Recommendation Algorithms Based On User Check-in Behavior

A lot of location-based services such as Location-Based Social Network(LBSN)have appeared with rapid development of the mobile Internet technology,which has attracted a large number of users.In the location-based social network,users can visit their favorite locations offline,check in their behavior in the location-based service,share interesting information with friends,and so on.The data in LBSN get an explosive growth with the booming of services in LBSN.There exist great commercial value and scientific value in analyzing users' history data,mining users' potential preference and recommending suitable point-of-interest for different users.At present,relevant researchers have put forward some models in point-of-interest(POI)recommending,mainly including matrix factorization models,Poisson factor models,link-based models and hybrid models.We made research on POI recommendation algorithm with methods based on graph link,which performs experiments in foursquare dataset.Finally,We put forward a hybrid algorithm for POI recommendation.The following is the main work of my research.Firstly,we introduced the background and significance in POI recommendation.We also introduced relevant research progress at home and abroad and algorithms based on graph link.Secondy,personalize PageRank is popular in calculating web pages' rank with users' personalized query.We use personalize PageRank to calculate other users'influence on a user in POI recommendation based on users' similarity.We use Bookmark-coloring Algorithm to realize effective calculation.That's our user model based on personalize Page Rank.Thirdly,considering the distance of POI,we user a power-law distribution to model the willingness of a user moving from one place to another as a function of their distance.PageRank is popular in calculating web pages' rank,and we use PageRank to calculate the popularity of POI.Then we can calculate the recommendation score of a user visiting POI on condition of his history POI set.That's our geographical model based on power-law distribution and PageRank.Finally,we have proposed a hybrid recommendation algorithm based on user model and geographical model.We use the linear weighted method to realize the hybrid mode.We have made experiments in common dataset to compare our algorithms with some classical algorithms,which proved our algorithm's promotion in POI recommendation.