|
|||||
|
|
| 一种词频与方差相结合的特征加权方法 | |
| 引用本文： | 邱云飞,王 威,刘大有,邵良杉.一种词频与方差相结合的特征加权方法[J].计算机应用研究,2012,29(6):2132-2134. |
| 作者姓名： | 邱云飞 王 威 刘大有 邵良杉 |
| 作者单位： | 1. 辽宁工程技术大学 软件学院,辽宁 葫芦岛 125105;吉林大学 计算机科学与技术学院,长春 130012 2. 辽宁工程技术大学 软件学院,辽宁 葫芦岛,125105 3. 吉林大学 计算机科学与技术学院,长春,130012 |
| 基金项目： | 国家自然科学基金资助项目（70971059）；辽宁省创新团队项目（2009T045） |
| 摘 要： | 通过分析特征词与类别间的相关性,提出了一种新的特征加权方法,依据特征词在特定类中出现的次数、特征词在某一类中的集中程度、特征词在特定类中的均匀分布程度来计算特征权值。通过与TF-IDF进行实验对比,新提出的TF-Var特征权重方法使得分类的微平均准确率得到了明显的提高。 |
| 关 键 词： | 文本分类 特征权重 词频 方差 |
Feature weighting method combined with word frequency and variance |
|
| QIU Yun-fei,WANG Wei,LIU Da-you,SHAO Liang-shan.Feature weighting method combined with word frequency and variance[J].Application Research of Computers,2012,29(6):2132-2134. | |
| Authors: | QIU Yun-fei WANG Wei LIU Da-you SHAO Liang-shan |
| Affiliation: | 1. School of Software Engineering, Liaoning Technical University, Huludao Liaoning 125105, China; 2. School of Computer Science & Technology, Jilin University, Changchun 130012, China |
| Abstract: | Through analyzing the defects of original feature weighting method, this paper proposed a new weighted method. This method calculated the feature weights based on the frequency of feature words appeared in the particular class, the concentration of feature words in one class, and the well distribution of feature words in the particulor class. Through the contrast of the TF-IDF feature weighting method, the new TF-Var feature weighting method has improved the micro-averaging precision of the classification obviously. |
| Keywords: | text classification feature weight word frequency variance |
| 本文献已被 CNKI 万方数据 等数据库收录！ | |
| 点击此处可从《计算机应用研究》浏览原始摘要信息 | |
| 点击此处可从《计算机应用研究》下载全文 | |