A Recommendation Trust Model Based on E-commerce Transactions Content-Similarity

Abstract

This paper proposes a new recommendation trust model, which is based on E-Commerce transaction content similarity and differentiates the trust degree of acquaintance node recommendation from stranger node recommendation (hereinafter referred to as TCSRTrust). The TCSRTrust model eliminates a subjective hypothesis that recommendation of a trustworthy node is the more trustworthy in previous global trust models, as the subjective hypothesis is not to conform to actualities in the current large-scale distributed network environment, and objectivity and reliability can not be guaranteed as a result. In contrast, simulation experiments prove that the TCSRTrust Model conforms better to the current new network application environment, and that the TCSRTrust Model brings greater improvement and enhancement in such broader security issues as fending off malicious node slanders and containing collaborative cheating.