Point-of-interest recommendation method based on geographic location fusion and category popularity

Abstract

The invention discloses a method for recommending a point of interest by matrix decomposition based on geographical location and social influence and category popularity, which obtains the historical check-in information of all users and candidate points of interest in the location social network, and calculates the geographic correlation coefficient of the target user to obtain The social network information of all users, calculate the social correlation coefficient S(x u,l ); Based on the preference of the candidate point of interest category and the popularity of the candidate point of interest, calculate the category popularity correlation coefficient C(y between the target user and the point of interest u,l ); through the probability matrix decomposition method, the geographical information model, social correlation coefficient, category popularity correlation coefficient and potential features are integrated to form a point of interest recommendation model, and the recommendation score Rec of the target user is calculated u,l ; Based on recommendation score Rec u,l Generate recommendation information, so as to achieve the purpose of personalized recommendation.

Description

technical field [0001] The present invention relates to a point of interest recommendation method, more specifically, to a matrix decomposition method for point of interest recommendation based on geographical location and fusion of social influence and category popularity. Background technique [0002] With the rapid development of social networks, smart mobile terminals and automatic positioning technology, location-based social network (Location-Based Social Network, LBSN) emerged as the times require, providing people with extremely convenient location services. Typical LBSN applications include Foursquare , Yelp and Gowalla etc. LBSN connects the online virtual society with the offline real world. Users visit the points of interest they are interested in offline. The check-in experience of the point of interest. POI recommendation is an important part of the LBSN service. By counting the historical check-in data of users, it aims to recommend potential POIs for users....

AI Technical Summary

Problems solved by technology

Therefore, the user-POI check-in matrix used in different collaborative filtering methods becomes very sparse, and the extreme sparsity of data will undoubtedly bring great difficulties to POI recommendation, which will lead to low accuracy of recommendation results

Images