Volume 3 - Issue 2
Web page automatic categorization based on non-linear SVM decision tree
Abstract
In this paper, a novel approach towards Web page categorization is proposed. First, we improve feature extraction and representation methods of web page. Compared with conventional methods, it can represent web page more comprehensively in content and structure. In the Multi-class classification, SVM is extended to non-linear SVM by using kernel functions, and the method of NSVM decision tree is presented based on traditional SVM decision tree. Finally, experimental results have shown that very high accuracy of classification has achieved by NSVM, and the recall is quite good. The presented method is effective and feasible.
Paper Details
PaperID: 42549170676
Author's Name: Wang, J., Yao, Y., Liu, Z.
Volume: Volume 3
Issues: Issue 2
Keywords: Character extraction, NSVM decision tree, Support vector machine (SVM), Vector space model (VSM)
Year: 2008
Month: April
Pages: 449-454