基于项目属性的用户聚类协同过滤推荐算法

Collaborative filtering recommendation algorithm based on user clustering of item attributes

协同过滤推荐算法是个性化推荐服务系统的关键技术,由于项目空间上用户评分数据的极端稀疏性,传统推荐系统中的用户相似度量算法开销较大并且无法保证项目推荐精度。通过对共同感兴趣的项目属性的相似用户进行聚类,构建了不同项目评价的用户相似性,设计了一种优化的协同过滤推荐算法。实验结果表明,该算法能够有效避免由于数据稀疏性带来的弊端,提高了系统的推荐质量。

Collaborative filtering recommendation algorithm is key technologies of personalized recommendation system, as the serious sparsity data of rated items, the similar users of active user is distribution of scattered. The traditional collaborative filtering recommender system algorithm consumes too many resources to search the nearest neighbor, as well as reliability is poor. A novel collaborative filtering recommender algorithm based on user clustering of item attributes is proposed. This algorithm reduces the negative effect on quality of recommendation, and shrinks the searching scope, which improves recommendation results for the system.