Based On Social Network Comprehensive Trust And Popularity Commodity Personalized Recommendation

Personalized recommendation system is now widely used in various fields,because it can solve the people in the network life found "information overload" problem,and to provide users with the required goods,services,is recommended.Interested in user modeling is the recommended the most critical part of the process,the stand or fall of user interest model is set up directly affect the recommendation results are accurate.Because the user's interest is often change,the traditional personalized recommendation USES the time window method and forgotten function method can't solve the problem of user happened new interest.In recent years,with the continuous development of network technology,the recommendation algorithm based on social network has become a hot spot in the field of personalized recommendation.Recommendation algorithm based on social network trust in acquiring the target users of adjacent matrix,consider only trust relationship between users,ignoring the similarity relation between users.To solve the problem of user interest degree of migration and the lack of trust algorithm based on social network,this paper's main work is as follows:First,the analysis of the existing time window method and forgotten function method to solve the user interest migration,presents a personalized recommendation method based on social network,the method to solve the user interest suddenly changes recommended by the traditional methods can not effectively.Second,the analysis of the existing trust personalized recommendation method based on social network,proposes a personalized recommendation method based on the comprehensive social network trust.Comprehensive trust is a kind of comprehensive considering the trust relationship between social network users with similarity measurement,it fully reflects the degree of close contact between users.Existing in the personalized recommendation algorithm based on social network trust tru st comprehensive trust,change to the adjacent matrix of the user and the target users as there is a trust relationship and have higher similarity relation.Comprehensive trust will be based on social network(comprehensive trust)recommendation algorithm and user evaluation score(UM)algorithm,the project evaluation score(IM)algorithm,and user-based collaborative filtering recommendation(UBCF)algorithm,based on collaborative filtering recommendation(IBCF)algorithm of the project,based on social network recommend trust models(RSTE)probability matrix decomposition algorithm experimental comparison,using standard quasi variance RSME as evaluation index,the experimental results show that using comprehensive trust recommendation algorithm on the recommendation accuracy improved.Third,recommend(CT)algorithm to produce the results the lack of diversity is less,the popularity of higher commodity into the recommendation system to produce the final recommendation results,based on social network comprehensive trust personalized recommendation algorithm to the popularity of goods(comprehensive trust and popularity of commodity,CT-PC)."Matthew effect" existing in the electronic commerce show fashion products will impact on the user's purchase behavior,can change the user's behavior habit and interest.Fashion products recommended by users join the recommendation system can influence buying behavior and to provide users with more variety of choices.Comprehensive trust will be based on social network and product popularity and personalized recommendation algorithm based on social network synthesis experiment comparison trust of personalized recommendation algorithm,using the recommended sample types as evaluation standard,the experimental results show that based on social network synthesis trust and popularity of recommendation algorithm to produce goods recommendation result has higher diversity.