区分用户长短期兴趣的IBCF改进算法

An Improved Item-based Collaborative Filtering Algorithm with Distinction between User's Long and Short Interests

协同过滤算法已被成功应用于许多领域,但遇到了可扩展性和精度低等问题,目前提出了许多改进算法,但它们均忽视了用户长短期兴趣对推荐的不同影响.针对这个问题,介绍了一种改进的长短期兴趣数据权重策略,它的关键是识别用户长期兴趣,为此提出了基于资源类别相似性和基于访问资源类别出现频率两种识别方法,并详细分析了这两种识别方法的优缺点.实验表明,将上述方法引入基于资源的协同过滤算法中,能提高推荐精度.

Collaborative filtering （CF） algorithms have been successfully used in many applications. But they have some problems, such as scalability, lower precision. And many improved CF algorithms have been proposed. However, all of them can＇t make a clear distinction between user＇s long and short interests for improving recommendation precision. In order to solve this problem, animproved weight strategy, consociating user＇s long and short interests data weighting method is introduced. The key of this method is how to identify user＇s long interests. So, two user＇s long interest recognition methods have been proposed： one is based on item＇s category similarity,the other is based on frequency for visited item＇s category. Additionally, advantages and disadvantages of these methods are analyzed in detail. The experimental results show that the proposed algorithm with two user＇s long interest recognition methods outperforms other itembased collaborative filtering algorithms.