提出了一种针对于电子商务的非集中式的信任管理模型。在该模型中,充分考虑电子商务的特点,较全面地引入影响信任度的信任因素,如时间因素、交易金额因素、邻居节点的信任度、邻居节点的规模以及对恶意行为的惩罚机制,解决了局部声誉和全局声誉的精确计算问题,提高了信任机制抗攻击的能力。对信任模型的交易成功率进行了仿真实验,与一些现存的信任模型相比较,新的模型能更有效地防止恶意行为。 Proposed a non-centralized trust management model for e-commerce. In this model, full consideration is given to the characteristics of e-commerce, more comprehensive introduction of trust factors that affect trust, such as time factor, transaction amount, neighbor node trust, neighbor node size and penalty mechanism for malicious behavior The accurate calculation of local reputation and global reputation raised the ability of the trust mechanism to resist attacks. The success rate of trust model was simulated experimentally. Compared with some existing trust models, the new model can prevent malicious behaviors more effectively.