Method of Text Feature Extraction Based on Relative Word-frequency
基于相对词频的文本特征抽取方法

Brings forward an improved model of evaluation function based on relative word-frequency, which is promised to re-solve above problem. This model is combined with the modern Chinese language word-frequency statistic achievements, extract feature words based on relative frequency, which results to better distinction degree between the defined classes and undefined classes. At last, the experiment has proved the feasibility and effect of this method.