Abstract: This paper presents a new recommendation method, which combines the similar relation in attributes and characters of items to user-based collaborative filtering recommendation algorithm by fuzzy clustering algorithm. The method transforms the users' preferences of single item to similar groups, which forms the dense preferences of users-fuzzy cluster. Then this method predicts item ratings that users have not rated by the similarity of items in similar groups. Finally this method realizes the user-based collaborative filtering recommendation algorithm based on the above steps. The experimental results show that this method can provide better recommendation results than traditional user-based collaborative filtering recommendation algorithm.

Key words: Recommender system; Collaborative filtering; Fuzzy clustering; Fuzzy cluster

摘要： 提出了一种运用模糊聚类方法将项目属性特征的相似性与协同过滤推荐算法相融合的推荐方法，此方法将用户对单个项目的偏好转化为对相似群组的偏好，目的是构造密集的用户－模糊簇的偏好信息，同时利用项目之间在相似群组的相似性来初步预测用户对未评价项目的评分，在此基础之上再完成基于用户的协同过滤推荐算法。实验结果表明，该方法确实可提高协同过滤推荐算法的推荐精度。

关键词: 推荐系统；协同过滤；模糊聚类；模糊簇