多主体系统的信任管理研究.pdfVIP

摘要 信任管理一直是一个比较热门的研究领域。目前,主体技术广泛应用于电 子商务。在基于主体的电子商务系统中,信任管理问题表现得更加明显。 本文研究多主体系统中的信任管理问题,主要完成了如下工作: 1. 结合信任管理本身和多主体系统的特点,分析并改进流行的信任计算模 型 D-S理论,提出了一个信任值评估模型。该评估模型引进恶意推荐监 测功能,通过欺诈因子对返回的推荐信誉进行选择,减少了恶意推荐带 来的影响;同时,引入了主观系数和时间因子,体现了信任计算过程的 动态性。 2. 研究多主体系统的特点以及多主体系统环境,借鉴现有多主体系统中的 联盟研究,提出层次化的信任管理模型。在层次化的信任管理模型中, 采用分组的方式对信任过程进行管理,分别考虑组内和组间的信任管 理,建立了层次化信任管理模型的体系结构。 3. 研究多主体系统模拟平台,介绍了实验平台的体系结构。设计了本文实 验的场景与方案,利用 Java 程序模拟了三种实验环境,同时实现了两 种信任管理模型思想。让模型在相应的环境下运行,分析本文提出的层 次化信任管理模型在效率、稳定性以及防欺诈能力上的表现。 关键词:信任管理;主体;多主体系统;层次化的信任管理模型;恶意推荐 Abstract Trust Management is a hotter research field. At present, agent technology is used widely in e-commerce. In the e-commerce systems which based-on multi-agent, the problems of trust management become more prominentIn this paper, we research trust management in multi-agent system, and have finished work as below: 1. By analyzing the popular D-S theory, and considering the feature of the multi-agent system, we improve and propose a trust evaluation model. Also in this model, we introduce the deception detecting function by the deception factor to selective consideration for returned referral information, and then reduce the impact of the deception. And we use objective factor to make an agent decide the percent of every dimension when calculating the trust value2. By studying the feature of the multi-agent system and the surrounding of the multi-agent system, and drawing on union research about the multi-agent system, we propose the hierarchical trust management modelIn this model, trust is managed by group. The procedure of trust management is implemented inside group or inter-group. Finally, we prompt the architecture of the hierarchical trust management3. We research some platforms of agent simulation, and introduce the architecture of JADE platform. Then we design a plan for simulation. In the simulation, we implement three simulation surroundings based on Java language, besides two trust management models. We run the trust