【摘要】 提出一个新型的面向分布式环境的信任评估模型:GTruMod。GTruMod基于后验概率思想分析实体间的协作历史,推导实体间直接信任,基于信任的社会模型特点计算实体的信誉,综合直接信任和信誉给出信任度的计算方法。基于该模型,该文提出一种基于图搜索方法对实体间信任度实施评估的算法,分析了该算法的复杂度。通过实验分析了该模型的评估性能特征,说明模型具有防范恶意推荐的能力。更多还原

【Abstract】 The paper proposes a novel decentralized trust evaluation model for grid, called GTruMod. In GTruMod, a formula is proposed toevaluate direct trust by referring to the theory of posteriori probabilities of binary events. Referred to characteristics of the society model about trustrelationship, methods are presented to discount feedbacks, compose recommendations, and evaluate the trust value. In decentralize networkenvironments, based on GTruMod, a practical graph–based search algorithm and some optimization methods are proposed to evaluate trust,furthermore its complexity is analyzed. The comprehensive experimental results prove that the model has ability to prevent the maliciousrecommendation.更多还原