基于用户反馈的智能合作过滤模型的研究  

Research of Agent collaborative filtering model based on user's feedback

作　　者：柯佳[1] 程显毅[2] 李晓薇[2] 

机构地区：[1]江苏大学工商管理学院,江苏镇江212013 [2]江苏大学计算机科学与通信工程学院,江苏镇江212013

出　　处：《智能系统学报》2007年第1期59-63,共5页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金资助项目(60473039).

摘　　要：为了提供给用户更准确的信息,提出基于用户反馈的智能合作过滤模型和一种基于用户兴趣的动态Q学习算法,并建立用户兴趣模型.通过隐式反馈和显式反馈这2种反馈方式更新用户模型并实现合作过滤.实验结果表明,在输入相同查询提问情况下ACFM在预测用户兴趣的效果和推荐搜索信息的查全率和查准率方面比传统的搜索引擎有明显改善.In order to serve users with more accurate information, the Agent collaborative filtering model- ACFM based on users＇ feedback and the dynamic Q learning algorithm are put forward, and users＇ interesting model is built. ACFM uses the method of users＇ interesting feedback consisted of implicit feedback and interactive feedback to realize collaborative filtering. Experimental results show that compared with traditional search engine, ACFM is more effective in predicting users＇ interests, and has more recalls and precision degree in recommending information when inputting the same inquire words.

关 键 词：合作过滤 AGENT 用户兴趣 Q学习 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]