一种用于电子商务的Web服务信任评估方法

摘要

为了减少不良行为对基于Web服务的电子商务的影响,提出了在信任评估中应遵循的基本原则,给出了全局评估方法的Web服务体系结构,并把适合这种结构的信任评估分为两个部分:第一个部分是业绩信任度,它受到交易结果、服务质量、交易时间和交易金额的影响；第二个部分是评价信任度,它除了受到影响业绩信任度那些因素外,还受到服务交易对象的评价、评价对象信任度以及防恶意评价等因素的影响.仿真表明:时间衰减因子取值越大,信任度变化越快；失败罚因子越大,当交易失败时信任度下降得越多；当出现大额欺诈行为后,信任度会降得很低；对团伙哄抬或压低服务信任度有抑制作用.与概率统计模型、平均模型和EigenRep模型相比,对突然不合作和周期摇摆行为更灵敏.此信任评估方法能对基于Web服务的电子商务中各种交易行为作出快速准确的评估.%To eliminate the effects on web-service-based e-commerce by bad behaviors, the basic guidelines in trust evaluation approach are presented. Web service architecture of global evaluation is proposed, and the trust evaluation for the structure is divided into two parts. The first part is business trust degree that is subject to trading result, quality of service, trade time and transaction amount. The second part is evaluation trust degree that is affected by not only factors affecting business trust degree but also assessments of trading object, trust degree of assessment object and preventing malicious evaluation factor. Simulations show that the greater the attenuation factor is, the faster the trust degree changes, and the bigger the failure penalty factor is, the more the trust declines when transactions fail. The evaluation approach can inhibit team cheat. Compared with probability statistical model, mean trust model and EigenRep model, this approach is more sensitive to the behaviors of suddenly no cooperation and swing. The trust evaluation method can assess various transactions fast in web-service-based e-commerce.