基于混合特征的中文文本分类

特征抽取是中文文本分类的重点和难点,文中比较了不同特征单元对分类性能的影响,将字特征与词特征相结合以期更好地表现文本特征。并在构建的实验系统中比较了不同特征单元的分类准确性,发现采用混合特征来进行分类,能得到较好的分类效果。

Chinese Text Categorization based on Mixed Features

Features extraction is keystone and difficulty of Chinese text categorization.This paper presents the influence of different feature units in text classification,and integrates the Chinese character feature with the world feature to efficiently represent the text character.An example system is designed to compare the classification accuracy of different feature units.The preferable categorization effectiveness is achieved by using mixed features to do the classification.