摘要
在Item-Based协同过滤算法中,项目之间相似性的度量是整个算法的关键.通过分析传统的相似性度量方法在系统评分数据稀疏的情况下所存在的弊端,提出一种新的用于Item-Based协同过滤算法的相似性度量方法,该方法从邻近度、影响力、有用性三个方面综合考虑了用户评分对项目相似性的影响.实验结果表明,该方法能够有效地避免传统相似性度量方法所存在的问题,使得数据稀疏性对最终推荐结果的负面影响变小,在一定程度上提高系统的推荐精度.
In Item-Based collaborative filtering algorithm, the most critical component of it is the similarity measure between items. This paper presents a new similarity measure applying to Item-Based collaborative filtering algorithm by analyzing the defects of traditional similarity measure in data sparsity. ,The measure takes synthetically into account the influence of user rating to item similarity from three aspects of proximity , effect and value. The experiment shows that the measure can avoid the defects of traditional similarity measure . Reducing the negative effect on the final recommendation and providing better recom- mendation results for the system.
出处
《小型微型计算机系统》
CSCD
北大核心
2009年第4期716-720,共5页
Journal of Chinese Computer Systems
基金
教育部科学技术研究重点项目(205014)资助
河北省教育厅科研计划项目(2006143)资助
