对等网信誉系统的一个关键问题就是信誉评估的准确性,一般通过考察推荐方的反馈可信度来给其反馈加权.提出一种新的反馈可信度评估模型,模型基于推荐方历史推荐行为满意度预测其未来的推荐行为,模型提出的“虚检验”机制加速了评估的过程,并且提高了准确性.仿真和分析表明,模型能有效抑制恶意推荐行为,降低其对最终信誉评估的影响.
In the Reputation Systems for Peer-to-Peer networks, a key issue is the accuracy of the evaluating of the reputation, a general method is concerned to weight the feedback by feedback trustworthiness of the recommender. A novel model for evaluating this trustworthiness is proposed in the article. The model can predict the future recommending behavior based on the recommender's history satisfaction degree of recommending behavior. A "virtual verify" mechanism is also used to speed up the evaluating process, and to improve its accuracy. The simulation and analysis indicates its efficiency in restraining malicious feedback, and through the restraining, the affect of malicious feedback to the final reputation evaluation is reduced.