The Research Of Personalized Recommendation Technology Combines The Trust Relationship In Social Network

In recent years,with the rapid development of online social networks,the amount of online records has increased dramatically.It has become more and more difficult for users to obtain the required information from the vast sea of data.In order to meet the demand of the user's personalized service,all kinds of recommendation systems are appearing constantly.The existing social networks recommendation methods only use the direct trust relationship and ignore the indirect trust relationship between users,the time sequence information and data sparseness problem,which make recommendation quality not high enough and can't meet the customers' requirements of filtering out items effectively.In view of the above problems,this paper focuses on the study of trust relationships among uses,time information's influence on the social networking recommendation algorithm.Based on this,two new recommendation algorithms are proposed to improve the users' satisfaction and optimize the personalized recommendation effect.The main work of this paper is as follows:(1)To solve the problems of the recommendations based on trust relationship in social networks,comprehensive considering the similarity between the users and the information such as direct trust and indirect trust relationship,this paper proposes an improved trust-aware recommendation approach.It integrates the similarity between the users and trust relationships into the process of probability matrix factorization,and carries out analysis of latent factor feature between the users trusted by the target user and the users who have the similar preference with the target users,to alleviate the problem of data sparseness.(2)In order to further improve the accuracy of recommendation algorithm in social networks,this paper takes the user-item rating matrix,the trust propagation mechanism between users,as well as the time sequence information into consideration and proposes an improved trust-aware aided time sequence recommendation approach.It puts trust relationships information between users and the time sequence information between items into probability matrix factorization model to learn the latent feature vectors both of users and items.It can better alleviate the cold start problem and improve recommendation accuracy.In addition,this paper adopts multi-dimensional evaluation indicators and comprehensively evaluates the proposed two social network recommendation methods based on the trust through extensive experiments.The experimental results show that compared with the existing recommendation methods based on social network,the proposed trust-based social network recommendation methods improve the accuracy of recommendation and have the obvious superiority.