As a machine intelligence paradigm, the support vector machines (SVMs) have tremendous potential for helping people to classify text document into a fixed number of predefined categories. The purpose of this paper is to discuss a new method of feature selection combined with principal component...