【摘要】 针对协同过滤推荐算法在网络个性化产品推荐服务中面临的覆盖率互斥、信息不对称问题,提出了一种融合用户动态标签和用户信任关系的改进协同过滤推荐算法,算法首先通过构建用户集、标签集和物品集三者之间的动态联系,建立用户动态偏好矩阵,接着构建基于用户社会网络信息的用户信任关系矩阵,以通过用户信任反馈机制实时更新用户间的信任值,最后通过融合用户动态标签和用户信任关系构建新的概率矩阵分解模型实现对协同过滤推荐算法的优化改进。仿真实验结果表明,改进算法提高了协同过滤推荐算法的性能,在一定程度上有效缓解了覆盖率互斥、信息不对称以及信任用户变窄问题对算法推荐质量的影响。更多还原

【Abstract】 To solve the problem of the mutual exclusion,asymmetric information and so on that the existing recommendation algorithms based on trust relation face in personalized recommendation service,an improved algorithm based on probability matrix factorization model which integrates user dynamic tags and trust relationships is proposed. In the proposed algorithm,the user dynamic preference matrix is firstly established by building the dynamic contact of the user sets,tag sets and item sets,and the user trust relationship matrix which uses the user trust feedback mechanism to update the trust value among users in real time is established based on user social network information,and then,finally,the matrix probabilistic factorization model integrating user dynamic tags and trust relationships is proposed to optimize collaborative filtering recommendation algorithms. The results of the simulation experiments show that the proposed algorithm improves the performance of collaborative filtering recommendation algorithm,and the above problems of mutual exclusion,asymmetric information and so on are effectively alleviated to some extent.更多还原