面向服务计算的信任预测模型

Trust Prediction Model for Service-oriented Computing

大多数信任预测模型的动态自适应能力较弱,且服务计算环境下代理之间交互的安全性较差。为此,提出一种面向服务计算的信任预测模型(SOC-TPM)。该模型结合人类认知行为,引入直接信任度、信誉推荐值、时间戳、历史交互记录等概念,通过创建动态信誉树对信誉关系进行建模,使信任预测模型更好地适应分布式计算环境。模拟实验结果表明,与J sang及Beth模型相比,该模型的信任预测准确度和平均相对误差分别提高了27%和47%。

Aiming at the lack of dynamic adaptive capacity of existing trust prediction model and security problem of the interaction between agents in services computing environment, this paper proposes a new trust prediction model in oriented-service computing combining human cognitive behavior. It introduces a direct trust value and reputation of the recommended values. In Trust Prediction Model for Service-oriented Computing（SOC-TPM）, it creates the credibility of the relationship model by a reputation tree and analysis the historical interaction records based on the timestamp to solve the problems of lacking dynamic adaptation capacity in traditional prediction model. Simulation results show that compared with J^sang model and Beth model, trust prediction accuracy and ARER of the model are respectively increased by 27% and 47%.