融合社会网络与信任度的个性化推荐方法研究

针对协同推荐技术存在的数据稀疏性和恶意评价行为等问题, 提出了一种新颖的基于社会网络的协同过滤推荐算法。该方法借助社会网络分析技术对协同推荐方法加以改进, 结合用户信任关系与用户自身兴趣, 通过计算网络节点的可信度来消减虚假评分或恶意评分给推荐系统带来的负面影响, 从而提高了推荐系统的准确度。实验表明, 相对于传统的协同过滤算法, 该算法可以有效缓解用户评分稀疏性及恶意评价行为带来的问题, 显著提高推荐系统的推荐质量。

Research on recommendation algorithm by fusing social network and trust

Aiming at data sparsity and malicious behavor in traditional collaborative filtering algorithm, this paper proposed a new algorithm of collaborative filtering based on socail network. To improve the accuracy of collaborative recommendation, this paper proposed a collaborative recommendation method based on social network analysis (SNA) by using SNA to improve the collaborative recommendation methods. This paper proposed a new social recommendation method combining user's trust and preference . The idea of this method was to compute the importance of the nodes to weak the negative influcen the false or malicious score bring to recommendation system. Experimental results show that the algorthm can alleviate the sparsity and maicious problems and achieve a better prediction accuracy than traditional collaborative filtering algorithms.