基于项目特征模型的协同过滤推荐算法  

A COLLABORATIVE FILTERING RECOMMENDATION ALGORITHM BASED ON THE MODEL OF ITEMS' FEATURES

作　　者：庄永龙[1] 

机构地区：[1]淮阴师范学院信息中心,江苏淮安223300

出　　处：《计算机应用与软件》2009年第5期244-246,共3页Computer Applications and Software

摘　　要：提出一种基于项目特征模型的协同过滤推荐算法。首先根据项目特征属性建立项目特征相似模型,在此模型基础上根据特征相似项目和用户评价相似项目,计算项目之间的综合相似度,弥补了以往协同过滤推荐算法在新项目推荐方面的不足。试验结果表明,该方法不但可以有效地改善传统协同过滤算法中新项目的冷启动问题,而且确实提高了推荐系统的推荐精度。A collaborative filtering recommendation algorithm based on the item features model is proposed in this paper. The similarity model of items＇ features was set up according to the feature attributes of items first, and then the comprehensive similarity was calculated between the items with similar features and with users evaluated similarities based on that model, so it makes up the drawback of the previous collaborative filtering recommendation algorithms in recommending the new items. The experimental results show that this method can not only improve the new items＇ ＂cold-start＂ problem in traditional collaborative filtering algorithms efficiently, but also soundly provides better recommendation precision in recommender system.

关 键 词：协同过滤推荐 冷启动 推荐系统 MAE 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论] TH162[自动化与计算机技术—计算机科学与技术]