| 中文文本分类中特征选择方法的比较 |
| |
| 引用本文： | 符发. 中文文本分类中特征选择方法的比较[J]. 现代计算机, 2008, 0(6) |
| |
| 作者姓名： | 符发 |
| |
| 作者单位： | 海南大学信息科学技术学院计算机系,海口570228 |
| |
| 摘 要： | 在自动文本分类系统中,特征选择是有效的降维数方法.通过实验对中文文本分类中的特征选择方法逐一进行测试研究,力图确定较优的中文文本分类特征选择方法.根据实验得出:在所测试的所有特征选择方法中,统计方法的分类性能最好,其次为信息增益(IG),交叉熵(CE)和文本证据权(WE)也取得了较好的效果,互信息(MI)较差.
|
| 关 键 词： | 特征选择 文本分类 向量空间模型 |
Comparison of Feature Selection in Chinese Text Categorization |
| |
| FU Fa. Comparison of Feature Selection in Chinese Text Categorization[J]. Modem Computer, 2008, 0(6) |
| |
| Authors: | FU Fa |
| |
| Affiliation: | FU Fa (School of Information Science , Technology,Hainan University,Haikou 570228) |
| |
| Abstract: | Feature selection is a valid method to reduce the dimension of text vector in automatic text categorization system.Through experiments on the feature selection methods in Chinese text categorization for testing research,trying to determine optimum feature selection methods in Chinese text categorization.According to the experiment that: all test of feature selection methods,the statistic has the best performance,followed by information gain(IG),cross entropy(CE) and weight of evidence(WE) also achieved good... |
| |
| Keywords: | Feature Selection Text Categorization Vector Space Model |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
