首页 | 本学科首页   官方微博 | 高级检索  
     检索      

三种文档语义倾向性识别方法的分析与比较*
引用本文:马海兵,刘永丹,王兰成,李荣陆.三种文档语义倾向性识别方法的分析与比较*[J].现代图书情报技术,2007,2(4):43-47.
作者姓名:马海兵  刘永丹  王兰成  李荣陆
作者单位:1. 南京政治学院上海分院,上海,200433
2. Autodesk公司,上海,200001
基金项目:中国博士后科学基金;江苏省博士后科学基金
摘    要:研究并实现三种文档倾向性识别的方法:基于情感词加权的方法、基于语义模式分析的方法和基于文本分类的方法。第一种方法利用特征词汇的情感语义倾向性。第二种方法对自然语言的句法结构进行简化,以获取合适粒度的倾向性语义模式。第三种方法则直接利用传统的基于文本分类的方法。通过在网络舆情分析系统中的具体实现,探讨这三种方法各自的不足和优势。

关 键 词:文本分类  倾向性识别
收稿时间:2007-03-06
修稿时间:2007-03-06

An Analysis and Comparison of Three Methods for Document Semantic Orientation Recognition
Ma Haibing,Liu Yongdan,Wang Lancheng,Li Ronglu.An Analysis and Comparison of Three Methods for Document Semantic Orientation Recognition[J].New Technology of Library and Information Service,2007,2(4):43-47.
Authors:Ma Haibing  Liu Yongdan  Wang Lancheng  Li Ronglu
Institution:1. Shanghai Branch, Nanfing Political Institute, Shanghai 200433, China;2. Autodesk Incorporation, Shanghai 200001, China
Abstract:This paper investigates and implements three methods for document semantic orientation recognition: one is based on the weighted emotional words; another is based on the analysis of the semantic pattern; and the third is based on text classification. The first method is try to use semantic orientation of feature words. The second method is to simplify the structure of the natural language syntax, in order to acquire the appropriate size patterns with semantics tendency. The third method is based on the direct use of traditional text classification method. Through experiments and the application in an analysis system of net-mediated public opinions, shortages and advantages of the three methods are proposed.
Keywords:Natural language process Text classification Semantic orientation recognition
本文献已被 维普 万方数据 等数据库收录!
点击此处可从《现代图书情报技术》浏览原始摘要信息
点击此处可从《现代图书情报技术》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号