协同过滤中的相似性度量方法的研究

现今,推荐系统越来越受到重视和普及,协同过滤算法是应用最为广泛的个性化推荐技术之一,对基于用户和项的协同过滤推荐算法进行简单的阐述之后,着重对相似性度量方法进行了研究,分别介绍了相关相似性、余弦相似性和调整的余弦相似性,在稀疏数据下对这3种相似性度量方法进行了分析与比较,在最终给出分析结论,并在此基础上提出了改进的相似性计算方法。

Research on Similarity Metrics for Collaborative Filtering

Nowadays,the recommendation system has been paid more and more attention and is more and more popular.The collaborative filtering algorithm is the most widely used personalized recommendation technology.After simply elaborating the user-based and the item-based collaborative filtering recommendation algorithms,this paper focuses on the similarity metrics including correlation similarity,cosine similarity and adjusted cosine similarity.Then,in the case of sparse data,it analyzes and compares these three similarity metrics.And the last,this paper comes to the final conclusion and proposes an improved similarity calculation method.