Multi-class SVM with negative data selection for Web page classification

Support vector machine (SVM) has been demonstrated its excellent performance in terms of solving document classification problem. In this paper, SVM with one-against-all structure is applied to solve Web page classification problems with multi-class. However, the main problem of SVM with one-against-all structure is that the negative data might be too huge so that the training time obviously increase. To solve this problem, a negative data selection method is presented to reduce a large amount of negative data for SVM. Experimental results show that the training time is obviously reduced. Moreover, the proposed method also keeps a high accuracy rate for Web page classification.