Abstract: Collaborative filtering is the main recommendation technology，which suffers from sparsity and scalability problems.A recommender system based on DSmTrust trust model is proposed.The trust transitivity is employed to overcome the sparsity problem and the distributed DSmTrust approach has good scalability.Experiments show that this study has better coverage than collaborative filtering and has higher accuracy than Massa’s trust-aware recommendation approach.

Key words: trust, recommender system, collaborative filtering

摘要： 协同过滤是当前主要的推荐技术，它的主要缺点是稀疏和扩展性问题。提出了一种基于DSmTrust信任模型的推荐系统，利用信任的传递性解决稀疏问题，分布式的DSmTrust方法具有良好的扩展性。实验表明，新方法比协同过滤的覆盖率更高，比Massa的信任感知推荐方法的精度更高。

关键词: 信任, 推荐系统, 协同过滤