基于主观Bayes方法和相关反馈技术的文档搜索重排序  

A Document Reranking Algorithm Based on Subjective Bayes Method and Relevance Feedback Technology

作　　者：赵滟[1] 顾益军[1] 

机构地区：[1]中国人民公安大学研究生院,北京100038

出　　处：《信息网络安全》2015年第4期41-44,共4页Netinfo Security

基　　金：公安部重点研究计划项目[2011ZDYJGADX016]

摘　　要：相关反馈方法作为查询扩展方法中的一种,已有向量空间模型中的Rocchio相关反馈算法、概率模型中的BIM(binary independence model,二值独立模型)、语言模型中的相关性模型(relevance model,RM)等算法模型。为进一步提高查询的准确率,文章提出一种结合主观Bayes方法和相关反馈技术的文档重排序算法,利用语言模型中RM3算法返回的扩展词项结合反馈文档集对文档重排序。实验中使用语言模型为基线方法,使用RM3方法为对比方法,通过与语言模型和RM3方法的比较表明本方法表现良好,在前N篇返回文档的正确率上优于语言模型和RM3方法。Relevance feedback method as a method of query expansion has many algorithm models, such as Rocchio relevance feedback algorithm of the Vector Space Model, BIM （binary independence model） of the Probabilistic Model, the RM （relevance model, RM） of the Language Modeling. In order to further improve the accuracy of query, this paper proposes a reordering algorithm combined with subjective Bayes method and relevance feedback technology. The documents are reranked by the extended terms and feedback documents. The extended terms are returned by RM3 algorithm. It used the Language Modeling as a baseline method, the RM3 method as the method of comparison in the experiments. It shows that our method performs well compars with the Language Modeling and RM3 method. The correct rate is better than the Language Modeling and RM3 method on precision at the first N returned documents.

关 键 词：主观BAYES方法 相关反馈 重排序 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]