文本特征选择方法的改进算法

传统的互信息特征选择方法受边缘概率的影响较大, 可能产生稀有词的概率评估分高于常用词的评估分, 从而导致倾向于选择低频词条的现象。为此，在分析了几种传统的特征提取方法基础上, 通过引入分散度及平均词频两个参数, 将互信息方法与特征的词频相关联, 从而使互信息的分类更加准确。实验结果表明, 该方法使分类效果更好。

Improved Feature Selection Method

Marginal probability has a greater effect on traditional mutual information feature selection method,which may leads to evaluation of r are words bigger than common words,resulting in selecting low frequency words. In order to improve these insufficiencies,we analyze several traditional featur e extraction methods,associates the mutual information method with characte ristics of word frequency by introducing disparity and average frequency,and in creases the accuracy of mutual information classification Experiment shows that this method makes better classification results.