基于信任关系的资源分配推荐算法改进研究

近年来，一些统计物理学的方法被用于推荐算法的研究中，其中，基于扩散的推荐算法研究成为一个重要方向。然而，这些方法都只关注用户对产品的评分信息，而忽略了用户之间普遍存在的信任关系。该文将用户信任关系引入到基于扩散的推荐算法中，提出了一种基于信任关系的资源分配推荐算法。该算法在资源分配的过程中，对受信任的用户用一个可调参数分配其更多的资源，从而提高受信任用户所选物品的资源值。在Epinions和FriendFeed两个真实数据集上的实验结果显示，该算法在准确性、多样性和新颖性等方面明显优于主流的基准推荐算法。

Improved Research on Resource-Allocation Recommendation Algorithm Based on Trust Relationship

In recent years, various recommendation methods have been proposed by referring to processes originated in statistical physics, among them the diffusion-based method is an important branch of study. However, these methods were proposed solely based on rating metrics, while the trust relations among users are always ignored. In this paper, we propose a novel information filtering algorithm by introducing users' social trust relationships into the original diffusion-based method based on the resource-allocation process. Specifically, a tunable parameter is used to scale the resources received by trusted users in the networked resource redistribution process. The objects collected by trusted users will receive more resources. Extensive experiments on the two real-world rating and trust datasets, Epinions and FriendFeed, suggest that the proposed algorithm has better performance than benchmark algorithms in terms of accuracy, diversity, and novelty in the recommendation.