关键词: 微博分类, 卡方统计量, 特征选择, 权值计算

Abstract: This paper analyzes the microblogging text feature information, and proposes a microblogging feature extraction method based on improved chi-square statistic. Firstly, the microblogging information classification features are expanded, microblogging features are increased frequency and other factors. It improves the traditional feature selection methods. Then, based on the traditional feature item weight calculation, the paper proposes a new improved method of CHI-square statistic for improving weight calculation results. Finally, the above method is tested by using the classical KNN and SVM algorithm, the experimental results show that this method improves the micro-blog information classification accuracy.

Key words: microblogging classification, CHI-square statistics, feature selection, weight calculation