Study of personalized trust model in enterprise computing environment

Abstract

The increasing popularity of resource exchange through peer-to-peer networks has encouraged the development of ways to support distributed and heterogeneous enterprise computing. Unfortunately, the prospect of distrust attracts agents seeking to weaken the network by propagating fraud and bad services, that is, establish the trust mechanism of the P2P network. Based on recommendation, a number of computational trust and reputation models were proposed for resource-sharing network based on agents. This paper presents a recommendation-based trust model, which utilizes theory of collaborative filtering and social network. Compared with existing trust model, this model has two main features. First, it eliminates malicious recommendation by evaluating credibility of recommender. Second, it adjusts calculating of path weight, while combining trust path, which addresses the problem of subjectivity in path weight initialization. Finally, comparing with Beth model based on eBay's data, the result shows the feasibility and effectiveness of this approach.