New approach to feature selection for text categorization using class correlation
基于类别相关的新文本特征提取方法

This paper proposed a new approach of feature selection for text categorization, which was based on the strong class correlation and positive class correlation, named SP. SP could eliminate the effect of negative and poor correlation feature effectly. SP discriminated between the positive feature and the negative feature by positive correlation factor, and eliminated the effect of negative feature. SP discriminated between the strong class correlation of features and the poor class correlation of features by positive class correlation factor, and eliminated the effect of poor correlation feature. SP could select high quality features effectively by combining these two factors. The result of Experiment indicates that the proposed approach has a good performance at categorization and reducing high dimensional feature space.