Abstract: The recommendation system plays a significant role in dealing with the information overload problem, but the recommendation system also has its disadvantages in terms of its data sparsity and cold start problems. Using traditional collaborative filtering algorithms can no longer satisfy the technical development of the recommendation system. With the development of social networks, the trust relationship of friends has been widely used in the recommendation system. However, in real life, the trust relationship in social networks also has the problem of sparse data. In order to better improve the quality of recommendations, a collaborative filtering recommendation algorithm for integrating potential social trust models is proposed. The new social trust model is mainly composed of the following parts：global trust values and expert models in the social matrix, an improved trust propagation model, an improved Pearson coefficient model. Through the analysis of the experimental results, it is known that the recommended algorithm for the fusion of the new model helps to improve the recommendation effect.

Key words: recommendation system, collaborative filtering, social network, trust model

摘要： 推荐系统在处理信息过载的问题上有着显著的作用，但是推荐系统也存在不足之处，在于它的数据稀疏性和冷启动问题，使用传统的协同过滤算法已经不能满足于推荐系统的技术发展。随着社交网络的发展，朋友信任关系被广泛地运用于推荐系统中。但是在实际生活中，社交网络中的信任关系也存在着数据稀疏的问题，为了更好地提高推荐的质量，提出了一种融合潜在社交信任模型的协同过滤推荐算法。新的社交信任模型主要由以下部分组成：社交矩阵中全局信任值和专家模型，改进的信任传播模型，改进的皮尔逊系数模型。通过实验结果分析可知融合新模型的推荐算法有助于提升推荐效果。

关键词: 推荐系统, 协同过滤, 社交网络, 信任模型