摘要
针对传统的Item-based协同过滤推荐算法在推荐系统应用中存在的不足,提出一种优化的Item-based协同过滤推荐算法.从项目相似性计算,项目近邻选取和预测评分计算三个方面对算法进行了优化,使计算结果更具有实际意义和准确性.实验结果表明,提出的算法可解决传统方法中由于数据稀疏所导致的相似性度量不准确的问题,并显著地提高了算法的推荐精度.
Although Item-based collaborative filtering recommendation algorithm is one of the most successful technologies in the recommendation systems,it still has such problem as poor recommendation quality.This paper presents an optimized Item-based collaborative filtering recommendation algorithm.In this paper,the calculation of similarity between items,the selection of neighbor items and prediction of ratings are optimized,which make the recommended result more meaningful and accurate.It can be proved that the optimized algorithm can solve the problem of the similarity measurement inaccuracy caused by the sparsity of data.The experiment results show that the algorithm is successful.
出处
《小型微型计算机系统》
CSCD
北大核心
2010年第12期2337-2342,共6页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60773198
60703111)资助
广东省自然科学基金项目(7300272
8151027501000021)资助
国家科技计划项目(2008X10005-013)资助
广东省科技计划项目(2008B050100040
2009A080207005
2009B090300450)资助
新世纪优秀人才支持计划项目(NCET-06-0727)资助
关键词
项目相似性
项目近邻选取
预测评分
Item-based协同过滤
推荐系统
similarity between items
selection of neighbor items
prediction of ratings
item-based collaborative filtering
recommendation system