Abstract: A Dirichlet trust rating model based on fuzzy theory is proposed to deal with the phenomenon of malicious evaluation and swing service in P2P network. The model proposed uses fuzzy clustering to isolate the malicious evaluation, and also reduces the probability of swing service. Considering the dynamic character of subjective trust, a new mechanism of trust renewal is provided, which is based on the degree of approximation in fuzzy theory. And this mechanism does better in improving the enthusiasm of service provider to provide quality services. Several related experiments show that the model has several good characters for the trust rating in P2P network.

Key words: fuzzy theory, Dirichlet distribution, malicious evaluation, swing service, trust rating

摘要： 针对P2P（Peer-to-Peer）网络中存在的恶意评价、摇摆服务等现象，提出一种基于模糊理论的Dirichlet信任评估模型。提出的模型利用模糊聚类的方式过滤掉一些恶意评价，并抑制节点摇摆服务的影响，从而为网络节点信任值的计算提供了公平、良好的评估环境。在此基础上，考虑到主观信任动态变化的特点，结合模糊理论中贴近度概念，提出一种新的模糊信任更新机制，从而更有效提高节点提供优质服务的积极性。实验结果表明，提出的模型具有良好的评估特性，适用于P2P网络信任评估的环境。

关键词: 模糊理论, Dirichlet分布, 恶意评价, 摇摆服务, 信任评估