推荐系统中一种新的相似性计算方法

A New Similarity Method Used in Recommender Systems

随着互联网的发展,推荐系统逐步得到广泛应用,协同过滤是其中的关键技术之一,它根据相似用户的喜好产生对目标用户的推荐。随着用户和项目数量的增加,用于产生推荐的数据集将极端稀疏,协同过滤系统的性能下降。为此,提出了一种新的用户多层相似性度量,不仅降低数据稀疏性的影响,而且克服了相似不相同的问题。实验表明,该度量方式能够提高协同过滤系统的推荐质量。

Recommender systems are becoming increasingly popular with the evolution of the Internet, and collaborative filtering （CF） is one of the most important technologies in recommender systems. The performance of CF systems degrades with increasing number of customers and items. So, a new multiple-level user similarity is presented, which not only overcomes the difficulty of data sparsity, but also solves the ＂similar but not same＂ problem. The experimental results show that the presented algorithm can improve the performance of CF systems in both the recommendation quality and efficiency.