Trust model has been widely applied to the field of P2P e-commerce, but the trust grain of the current trust model of P2P e-commerce is too rough to embody the peer's real behavior very well.In order to solve the problem, a novel fine-grain trust model (FG-trust) is presented, which can complete the trust computation of one peer in different domains and different aspects. The model adopts the technology of multi-agent, uses the agent to manage and maintain the network and divides the network into multiple domains. Each domain sets a domain-agent to manage the peers in it and a manager-agent is set in the entire network to manage the messaging between inter-domains. In order to measure the trust value of one peer in any domain and take full account of the impact of different relationships between domains on recommendation trust, the concept of domain-model is introduced into FG-trust. The model uses Bayesian network to compute the peer's trust value, which can predict the trust value of one peer in any aspects. And an updating method of Bayesian network model and domain-model is proposed, which can constrain the peer's malicious behavior effectively. Finally, the effectiveness and feasibility of this model are illustrated by the experiments.%针对现有P2P电子商务模型信任粒度比较粗糙,不能很好体现节点真实行为的问题.提出一种细粒度信任模型(FG-trust),可以计算一个节点在不同领域、不向方面的可信度.模型采用多代理结构,将网络划分为多个域,每个域内设一个代理,管理域内节点信息,整个网络中设置一个总代理管理域间消息传递.引入领域模型的概念,计算同一节点在不同领域的可信度,并考虑领域间相关程度对推荐可信度的影响.在可信度计算方面采用贝叶斯网络,可以有效预测节点在不同方面的可信度.并提出了贝叶斯网络和领域模型的更新方法,有效遏制了节点的恶意行为.最后通过仿真实验验证了模型的可行性和有效性.
展开▼
