本文已被：浏览 2103次 下载 3599次
Received:March 17, 2011 Revised:April 11, 2011
Received:March 17, 2011 Revised:April 11, 2011
中文摘要: 针对传统协同过滤推荐算法中以稀疏评分计算用户相似性可能并不准确的问题,提出以用户行为对应一定分值填补空缺的I-U 评分矩阵,并以分角色下的权重系数K 约束用户相似性计算的改进协同过滤推荐算法。实验表明,改进算法的推荐质量更高。
Abstract:There are sparse ratings problem in the traditional CF recommendation algorithm, and based on this sparse ratings will lead to the fact that the similarity may not be accurate. For this reason, a CF algorithm based on fixed I-U ratings matrix, which is given by a certain ratings of user behavior instead of vacancies rating, and weighted coefficient K bases on users’ role to constrain the similarity calculation is proposed. Experiments show that the improved algorithm has better recommendation quality.
keywords: collaborative filtering I-U score matrix similarity calculation users’ behavior users’ role.
文章编号： 中图分类号： 文献标志码：
基金项目:
引用文本：
李幼平,尹柱平.基于用户行为与角色的协同过滤推荐算法.计算机系统应用,2011,20(11):103-106
LI You-Ping,YIN Zhu-Ping.Collaborative Filtering Recommendation Algorithm Based on Users’ Behavior and Roles.COMPUTER SYSTEMS APPLICATIONS,2011,20(11):103-106
李幼平,尹柱平.基于用户行为与角色的协同过滤推荐算法.计算机系统应用,2011,20(11):103-106
LI You-Ping,YIN Zhu-Ping.Collaborative Filtering Recommendation Algorithm Based on Users’ Behavior and Roles.COMPUTER SYSTEMS APPLICATIONS,2011,20(11):103-106
