检索
E-mail
RSS
作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2013, Vol. 39 ›› Issue (1): 175-178.doi: 10.3969/j.issn.1000-3428.2013.01.037

• 人工智能及识别技术 • 上一篇    下一篇

基于评分矩阵预填充的协同过滤算法178

彭 石，周志彬，王国军   

  1. (中南大学信息科学与工程学院，长沙 410083)
  • 收稿日期:2012-04-09 修回日期:2012-05-07 出版日期:2013-01-15 发布日期:2013-01-13
  • 作者简介:彭 石(1985－)，男，硕士研究生，主研方向：智能推荐；周志彬，博士研究生；王国军，教授、博士生导师
  • 基金项目:
    国家自然科学基金资助项目(61073037)

Collaborative Filtering Algorithm Based on Rating Matrix Pre-filling

PENG Shi, ZHOU Zhi-bin, WANG Guo-jun   

  1. (School of Information Science and Engineering, Central South University, Changsha 410083, China)
  • Received:2012-04-09 Revised:2012-05-07 Online:2013-01-15 Published:2013-01-13

PDF

1414

摘要： 随着用户和项目数量的增长，用户-项目评分矩阵变得极其稀疏，导致基于相似度计算的推荐算法精度降低。为此，提出一种基于加权Jaccard系数的综合项目相似度度量方法，使用项目综合相似度对评分矩阵进行预填充。实验结果表明，在用户-项目评分矩阵极其稀疏的情况下，该算法能产生比传统算法更精确的推荐结果。

关键词: 推荐算法, 协同过滤, 相似度, 信息熵, 加权Jaccard系数

Abstract: When the magnitudes of users and commodities grow rapidly, the rating matrix becomes extremely sparse. In the condition, algorithms based on traditional similarity computing have poor performance. In order to overcome this problem, this paper proposes a comprehensive item similarity measurement algorithm based on weighted Jaccard index, and prefills the rating matrix by the comprehensive item similarity. Experimental results show that the algorithm is more accurate compared with traditional algorithms.

Key words: recommendation algorithm, collaborative filtering, similarity, information entropy, weighted Jaccard coefficient

中图分类号: