Trust And Its Application In The Recommendation System

Along with the development of network technology, More and more people payattention to Information Overload and Information Security. The expanding demand inproviding user information that needed rapidly result in a more serious requirement onsecurity mechanism and recommender system. For security mechanism, previous studiesused to gain a reliable trust value by computing user data through different ways. Insteadof that, this paper attempt to fractionizes the trust and then model them. For recommendersystem, traditional studies used to investigate it as a whole, but being a whole may ignoreuser’s unique property. Through combining local structures in social network with therecommender system, more personal recommendation can be given.In this paper, trust mechanism which reflects the user’s inter-relationship anddependency relationship was studied through the information spreading in multi-agentsystem. Discipline of information spreading was investigated by fractionize trust. Firstly,objective trust and subjective trust were defined particularly and several characteristics ofboth were given respectively, closure definition and construction method are both givenlater on. Finally transitive trust was defined based on objective trust and subjective trust,several application models in recommender system were discussed in the end.For recommender system, the definition of interest sets was given, then based onthis definition the influence of the local network structure on recommender system wasproposed and proved both in theory and experiment. In the end an experiment was givenwhich combining one local property with the classical recommender system shows thepositive effect of the local structure on accuracy prediction.