基于领域本体的文本过滤模型  

Text filtering model based on domain ontology

作　　者：李嘉俊[1] 郑宇[1] 吴耿锋[1] 

机构地区：[1]上海大学计算机工程与科学学院,上海200072

出　　处：《计算机工程与设计》2008年第21期5555-5558,共4页Computer Engineering and Design

摘　　要：目前广为应用的文本过滤技术是利用关键字检索,没有考虑概念之间的关联,因此其过滤性能在达到一定程度后,很难有突破。介绍了一种基于领域本体的文本过滤模型DOTFM,探讨了领域本体在文本过滤中的应用。DOTFM在文本向量的表示和用户模板建立中引入概念关联度,并提出局部型和全局型的文本向量和用户模板。实验结果表明,DOTFM的召回率比之传统的基于关键字的过滤模型有较大提高,而其准确率在合适的阈值时,也有较大提高。The keyword based index is widely used in text filtering, which fails to deal with the relationship between concepts. Consequently, when the filtering performance reaches certain degree, it is very difficult to make a breakthrough. This paper introduces a text filtering model called DOTFM, and studies the applications of domain ontology in text filtering. In DOTFM, the concept related degree is introduced as a factor in text vector presentation and user model construction, and the local/global text vector and local/global user model are also proposed. Comparedwiththerecall-rateandprecision-rate of traditional key word based text filtering model, experimental results show that the recall-rate of DOTFM is improved significantly, and its precision-rate is also improved obviously under some proper thresholds.

关 键 词：领域本体 概念 关联程度 文本向量 文本过滤 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]