基于用户聚类和Logistic函数改进的协同过滤算法

Improved collaborative filtering algorithm based on user clustering and Logistic function

当前,互联网上存在的商品数以亿计,如何向用户推荐其感兴趣的项目成为一个极具挑战的问题。传统基于用户的协同过滤算法(collaborative filtering)在海量数据情况下存在推荐精度不高、多样性和新颖性不足的缺点。针对以上不足,提出一种基于用户聚类和Logistic函数改进的协同过滤算法。算法基于用户模糊聚类,通过融入用户特征属性相似度度量策略和Logistic改进的协同过滤算法来提升推荐效果。实验结果表明该算法能在保证推荐的有效性同时较好地兼顾推荐的准确性和多样性。

How to recommend a commodity to a user is a problem when its number is huge. Traditional collaborative filtering algorithm in accuracy is not high and lack of diversity and novelty in the big data environment. This paper proposed a improved collaborative filtering algorithm based on user clustering and Logistic function which improved collaborative filtering algorithm and incorporate user＇s features and attributes based on user-clustering. The experiment result shows that this algorithm is effective,at the same time it can ensure the recommended accuracy and diversity.