Identification of prevention against strategic cheating and fraud behavior in peer-to-peer environments are still an open issue. This paper proposed a reputation model PTMTrust which placed trust in mainstream view. Following the hypothesis that honest feedbacks were the mainstream, this model constructed a sliding window to move along the evaluations and regarded the clustering center of honest evaluations to coincide with that of the highest-density window. With fitting degree of evaluations, active degree of interacts and transaction number employed to measure credibility of evaluations, global trusts of peers properly reflect the quality of transaction and service, and could be used to punish cheating behavior and to restrain disturbance from dishonest evaluations. Therefore fairness of P2P system as well as successful transaction rate was improved. Hie simulation experiments verified the protection performance of PTMTrust. Compared with traditional models, due to more reasonable weights assigned to evaluations, the model weakens the threat from malicious behavior like conspicuous fraud effectively and curbs disturbance of dishonest recommendation.%针对P2P环境中策略性欺骗和作弊行为识别及防范难的问题,提出一种基于置信于主流意见的信誉模型(PTMTrust).基于诚实反馈是节点间交互主流的原则,采用最大密度方法,构建滑动窗口定位主流评价值聚类中心.用评价值拟合度、交互活跃度和交易数等参数,确定评价值的可信度,使全局信任度真实反映节点服务或交易质量,惩罚虚假交易行为,抑制不真实评价的影响,增强评价系统的公平性,提高交易成功率.仿真实验验证了模型的保护功能,与传统信誉模型相比,由于推荐权重得到合理分配,有效地缓解了合谋作弊等恶意行为的影响,遇制了不诚实推荐对信任模型的破坏.
展开▼
