Research On Personalized Recommendation Algorithm In The LBSN

The popularity of location technology and mobile Internet has led to the rise of location-based social networks.Friends and venues recommendation are important service content of location-based social networks.At the same time,personalized recommendation algorithms based on LBSN have become the focus of academic research.Based on LBSN,this article builds a recommendation algorithm for friends and venues to further improve the effectiveness of recommendations.In the aspect of friend recommendation,build a friend recommendation algorithm combined with online relations and line behavior by considering the location preference similarity,distance similarity and familiarity.First,calculate the similarity of the location preference by considering the time factor and the elimination time factor.Secondly,the distance similarity is calculated by exploring the distance of the checkpoint between the user and his friends.Third,using rank and route as important factors affecting the friendship relationship to calculate the familiarity;Finally,the above three features are weighted to calculate the final recommendation score.In terms of venue recommendation,analyze the disadvantages of venue recommendation algorithm based on bipartite graphs,and it is improved by using the trust relationship and distance two features.The degree of trust is calculated based on a random walk algorithm and user influence,and the distance feature is calculated using the number of points within 1 km from the check-in point of the target user.Using the data on Gowalla,this paper designs contrast experiments for friend recommendation algorithm and venue recommendation algorithm respectively.Three indicators including precision,recall,and F1-measure were selected to evaluate the algorithm.The results show that the two algorithms proposed in this paper achieves good performance in different evaluating indicators.