Point Of Interest Recommendation Algorithm Combined With Various Influence Factors

The popularity of mobile Internet devices has greatly promoted the development of location-based social networks(LBSN).Users can use LBSN to share the location experience in real time when they visit the point of interest.By analyzing the interactive data between users and their interest points,personalized recommendation of interest points can be made for users.In view of the problems existing in the research on the algorithm of interest point recommendation,such as data sparsity,insufficient information mining and low recommendation accuracy,this paper proposes the corresponding solutions by integrating various factors affecting interest point such as time and geography.This paper first analyzes the research status of the algorithm of point of interest recommendation.Aiming at the problem that the time information mining of interest point recommendation is insufficient,the time non-uniformity and time continuity of user behavior are analyzed,and the time pattern of user behavior is explored.This paper analyzes the visits of interest points in different time periods,introduces the time-popularity characteristics of interest points,combines them into the recommendation of interest points,and proposes the algorithm of recommendation of interest points that incorporates the time influencing factors.Secondly,in view of the data sparsity problem existing in the recommendation of interest points,this paper considers the social factors and geographical location factors of users,learns the personal preferences of users,generates the recommendation list of interest points,and expands the original check-in matrix of users with the interest points in the table to improve the matrix density.Thirdly,for the problem of low recommendation accuracy,this paper combines social influence factors and geographic influence factors into the recommendation of interest points,and integrates the geographical influence factors into the original matrix decomposition method in a predictive manner,and takes social factors as regular constraints.Optimize the loss function of the matrix decomposition.Then,the time influence factors are integrated,and the interest point recommendation algorithm that integrates multiple influencing factors is proposed.Finally,the two algorithms proposed in this paper are experimentally verified on real data sets.The results show that the two algorithms can effectively improve the accuracy of the recommendation results.