Research On Trust Incorporated Recommendation Of Point Of Interest

With the rapid development of Location Based Social Networks,massive amounts of user behavior information have been accumulated.In this context,how to accurately help users find interesting locations in massive information and reduce user decision-making time is an urgent problem in academia and industry.Point of Interest recommendation systems have emerged.An important goal of the Point of Interest recommendation system is the characterization of user behavior.However,existing Point of Interest recommendation methods often characterize user behaviors based on the user's preference for Points of Interest,resulting in a recommendation effect that is less than expected.The trust between user and Points of Interest reflects the user's interest and recognition of the Point of Interest,and influences the Point of Interest recommendation system to characterize user behavior.Therefore,this thesis introduces the trust calculation between users and Point of Interest in the Location Based Social Networks,and proposes and studies a method of Point of Interest recommendation incorporating trust.The main work and contributions are as follows(1)A method for obtaining trust between users and Points of Interest is proposed.Based on the principle of trust transitivity,from the social behavior of users,the trust between users and the check-in behavior information of users is used to build trust between users and Points of Interest(2)The way in which trust is incorporated into the recommendation process is studied According to the stages of integrating trust into the Point of Interest recommendation process,a comparative analysis of trust integration methods from the perspectives of pre-filtering,post-filtering,and modeling provides new ideas for researching the process of integrating trust Point of Interest.(3)The method of Point of Interest recommendation incorporated trust is achieved.Based on the theory of trust post-filtering,using the trust between the user and the Point of Interest,the Point of Interest prediction rating based on the user's sign-in behavior is optimized and adjusted,and the recommendation result that satisfies both the user and the interest point preference and the trust between the users and the Points of Interest,is obtained.(4)The Gowalla data set is used for experimental exploration of this thesis.And the precision,recall and F-measure was selected as indicators to measure the quality of recommendation.After obtained the optimal parameters of the recommendation method,conducts a comprehensive analysis of multiple sets of comparative experiments,and measures the recommendation effect of the proposed trust transitivity and trust post-filtering methods.The research shows that the method of obtaining trust between users and Point of interest based on trust transmission can accurately fit user behavior.And the proposed method of trust incorporated recommendation of Point of Interest is superior to the method of Point of Interest recommendation method based on user sign-in behavior.The results of this thesis have important theoretical and temporal significance for related research on improving the Point of Interest recommendation service.