基于用户行为特征分析的隐性信任协同过滤推荐方法

An Implicit Trust Based Collaborative Filtering Recommendation Approach by Users＇ Behaviors Analysis

个性化推荐技术能够根据用户评价项目的信息分析用户的喜好，采用信息的“推”技术为用户决策推送各种有价值的情报。传统的基于用户的协同过滤推荐方法在项目评分矩阵高度稀疏时推荐效果较差。基于用户信任关系的推荐方法，可回避评分数据不足的问题，但信任信息的获取存在诸多障碍。因此，本文提出了一种基于隐性信任的协同过滤推荐方法，通过对用户评分行为的分析建立用户对其邻居和对项目的隐性信任模型，根据信任邻居的历史喜好和用户自己的历史喜好向用户推荐感兴趣的项目。实验结果表明本文提出的方法能够为用户提供更准确的推荐结果。

Personalized recommendation technique can analyze users＇ preferences according to the information about rated items, and apply the ＂push＂ technique to provide a variety of valuable information for users to make decisions. Traditional user-based collaborative filtering recommendation approach might make poor recommendations when the rating matrix is highly sparse. Trust-based recommendation method can avoid the problem of lacking rating data, but there are many obstacles to obtain the so-called trust information. Therefore, in this paper, we propose an implicit trust based collaborative filtering recommendation approach, which builds two kinds of implicit trust recommendation models, i. e. , user-trust-neighbor and user-trust-item models, by analyzing users＇ rating behaviors. According to neighbors＇ and users＇ historical preferences, some interesting items can be recommended to the target users. The experimental results demonstrate that our proposed method can provide users with more accurate recommendations.