| 一种基于PCA和RS的文本特征抽取方法 |
| |
| 引用本文： | 康涛.一种基于PCA和RS的文本特征抽取方法[J].现代电子技术,2007,30(10):88-90. |
| |
| 作者姓名： | 康涛 |
| |
| 作者单位： | 信息工程学院,河南,郑州,450002 |
| |
| 摘 要： | 提出一种基于PCA(主成分分析)和RS(粗糙集)的文本特征抽取方法。首先利用PCA将n维词语特征-文档矩阵变换为一个m维的正交矩阵,再采用RS的方法对m维新特征进行进一步的约简。实验结果表明,新的特征抽取方法用于垃圾邮件过滤能有效的提高垃圾邮件过滤的正确率和召回率。
|
| 关 键 词： | 特征抽取 特征约简 |
| 文章编号： | 1004-373X（2007）10-088-03 |
| 收稿时间： | 2006-09-19 |
| 修稿时间： | 2006-09-19 |
A Method for Text Feature Extraction Based on PCA and RS |
| |
| KANG Tao.A Method for Text Feature Extraction Based on PCA and RS[J].Modern Electronic Technique,2007,30(10):88-90. |
| |
| Authors: | KANG Tao |
| |
| Affiliation: | Information Engineering College,Zhengzhou,450002,China |
| |
| Abstract: | The paper proposes a new method for text feature extraction based on Rough Sets theory and PCA.First,using PCA method transform a n-dimensional terms-documents matrix into m-dimensional orthogonal matrix.Then,a rough set method to the result of principal components analysis is used for feature reduction.Results show that using our method as the basis for an email classifier to filter out spam enjoys a very high degree of precision and recall. |
| |
| Keywords: | PCA RS |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
