Trust Calculation Model Based on Social Network and Evidence Theory

摘要

To solve the trust issue caused by malicious users in e-commerce, we propose a trust calculation model based on social network. Through analysing the trust features, social network is introduced into this trust model to get information from the most trusted source according to the subjectivity and the compositionality feature of trust to improve the buyers' situation of asymmetric information in electronic trading. In view of the advantage that conflicted evidences can be well combined by evidence theory, we resolve the conflict of information from each recommended third-party. We also introduce other various trust influence factors, such as transaction rating dictionary, transaction background important factor, transaction amount factor and trading time factor, etc. Finally, through simulation experiments, we set different confidence factors depending on different trading environment, and verify that the method to combine social network and evidence theory for obtaining a reasonable recommendation trust value is feasible. It proves that our proposed trust calculation model based on social network has a better anti-attack ability.