Researches And Implementation On Social Network And Trust Model Based Recommender System

Trust based recommendation technology has been one of the focuses in the recent field of recommender system. In order to describe the relationship of trust more precisely and improve the accuracy of recommendation, this paper mainly researched a trust model on the basis of social network and rating information of users and another recommendation model on the basis of trust network, improved random walk strategy and the introduction of domain authority of users.Main components of this paper have been listed as follows:(1) A trust model SRNT based on social network and rating information of users was put forward. A calculation method of direct trust degree was designed by combining users’ interaction strength in different domains with other users’reputation in related domains, as well as the method of trust transitivity in the path of trust.(2) Random walk of recommendation model TrustWalker was improved by the introduction of trust model SRNT. SRNT-ReferentialUserWalker, a recommendation model, based on trust network and improved random walk strategy was designed. The accuracy and coverage were improved compared with TrustWalker model.(3) The process of rating prediction of SRNTA-ReferentialUserWalker model was improved by the application of domain authority, by which, the accuracy of trust was further increased.(4) A prototype recommender system based on the recommendation model SRNTA-ReferentialUserWalker was designed and realized.Main achievements of this paper were also listed as follows:(1) A calculation method of direct trust degree which combines users’ interaction strength in different domains with other users’reputation in related domains.(2) A recommendation model SRNTA-ReferentialUserWalker based on trust network, improved random walk strategy and the introduction of domain authority of users.The research of this paper has a certain reference value to the recommendation algorithm based on trust.