Abstract: Transferring trust using its own social nature can effectively alleviate the data sparsity problem, and improve coverage and accuracy of recommender systems. While current researches on the trust network suffer some problems, such as inaccurate model construction, complicated trust transfer mechanism and distortion, etc. In order to accurately portray the trust relationship between customers in trust networks, this paper introduces the concept of support-trust value to establish a new customer trust model combining trust value with support-trust value, and then makes transfer rules from trust sociality. After that, this paper constructs a customer trust network based on this model and designs the corresponding personalized recommendation algorithms. The experimental results show that this model can improve coverage, accuracy and recommendation quality of the recommendation system.

Key words: E-commerce, customer trust network, trust value, support-trust value

摘要： 利用信任的社会性质进行信任传递，可有效缓解数据稀疏的问题，提高推荐系统的覆盖率和准确率。目前对信任网络的研究存在信任模型建立不准确、信任传递机制复杂与失真等问题。为准确表述信任网络中的客户信任关系，引入信任支持度的概念，提出了一种信任度与信任支持度相结合的客户信任模型；制定了符合信任社会性的传递规则，构建了基于该模型的客户信任网络，并设计了相应的个性化推荐算法。实验结果表明，此模型提高了推荐系统的覆盖率、准确率及推荐质量。

关键词: 电子商务, 客户信任网络, 信任度, 信任支持度