Max-difference maximization criterion: a feature selection method for text categorization

摘要

1 IntroductionText categorization methods based on machine learning always encounter the curse of dimensionality. Therefore, it is crucial to perform feature selection to reduce dimensionality of text vectors. Methods based on the difference between document rate of a term in the positive class and that in the negative class have been widely studied in recent years, which is originated from balanced accuracy measures (ACC2) [1].