基于用户模糊聚类的综合信任推荐算法

针对传统协同过滤推荐算法通常存在的数据稀疏和冷启动问题，根据用户间的信任关系，提出基于模糊C均值聚类的综合信任推荐算法。采用评分数据和信任数据计算用户间的隐式信任值和显式信任值，利用显隐式信任得到综合直接信任值，基于信任的传递特性获得Jaccard全局信任值，最终通过动态结合综合直接信任与Jaccard全局信任获取综合信任值，同时将信任机制融入模糊C均值聚类算法实现对目标用户的精准推荐。在FilmTrust真实数据集上的实验结果表明，该算法有效缓解了数据稀疏和冷启动问题，并且相比传统协同过滤推荐算法具有更高的推荐质量。

Comprehensive Trust Recommendation Algorithm Based on User Fuzzy Clustering

The traditional collaborative filtering recommendation algorithm is limited by data scarcity and the cold start problem.Leveraging the trust relationships between users, this paper proposes a comprehensive trust recommendation algorithm based on Fuzzy C-Means(FCM) clustering.The algorithm employs the rating data and trust data to obtain the implicit trust value and explicit trust value between users.On this basis, the comprehensive direct trust value is calculated.Then based on the transmission characteristics of trust, the global trust value of Jaccard is acquired.Finally, the direct trust value and Jaccard global trust value is dynamically fused to obtain a comprehensive trust value, and the trust mechanism is integrated into the FCM clustering algorithm to implement accurate recommendation for target users.Experimental results on the real dataset FilmTrust show that the proposed algorithm effectively solves the problem of data scarcity and cold start, providing better recommendation performance than the traditional collaborative filtering recommendation algorithms.