Research On Point-of-interest Recommendation Algorithm Based On Incremental Updating

In recent years,with the rapid rise of Location-Based Social Network(LBSN),point-of-interest(POI)recommendation has become an important application method.It is designed to recommend interesting but unvisited locations to users based on their historical check-ins and other supporting information.However,due to the continuous influx of a large number of new users and new interest point-of-interests in the current LBSN platform,the increasingly serious information overload problem starts to be exposed to users.In order to recommend relevant technologies based on point-of-interests to have an effect on this problem,this thesis mainly conducts improvement research from the following two aspects:1.User check-in data contains information of multi-source situational factors,such as geographical location,time,social relationship,emotional tendency,text information,etc.Different situational factors can reflect different aspects of users' behavior and habit,using these information in a reasonable and comprehensive way can effectively improve the quality of recommendation.In view of the insufficiency of the existing LBSN research on the mining of user semantic location information related to the point-ofinterests,and the fact that most POI recommendation algorithms ignore the influence of the region where point-of-interest is located on the recommendation results,this thesis proposes a point-of-interest recommendation algorithm(USTTGD)based on semantic location and regional division.By integrating various situational factors into a unified recommendation framework,the algorithm achieves better results on Foursquare and Gowalla datasets,especially on the recommendation accuracy.2.For the current point-of-interest recommendation system,with the continuous accumulation and increase of user check-in data flow,the computational complexity of the algorithm significantly increases,which will affect the real-time performance of the system.This thesis proposes an incremental update mechanism based on POI similarity,under which USTTGD algorithm is implemented,and a point-of-interest recommendation algorithm(IUUSTTGD)based on incremental update is obtained.The invariants decomposed in the calculation process of the similarity of point-of-interest and the dependent variables changing with the check-in data are used,and only these changes are changed when the system updates the data.Through experimental comparison and analysis on real datasets,it can be seen that the incremental update mechanism based on POI similarity has obvious advantages in recommendation efficiency,which can update the recommendation results in real time and effectively improve the scalability of the system under the condition of ensuring high recommendation accuracy.