Application of SVM-KNN Classifier into Web Page Classification

The SVM-KNN method is applied to Chinese web page classification in order to improve the classify accuracy of Chinese Web page. After analyzing the characteristics of web pages,a new vector representation of web pages was proposed and applies to web page classification. A multi-class classifier is constructed using SVM after the downloaded web pages was presented as a vector of vector space.In the classify phase,the distance is computed from vector to hyper-plane,based on which the SVM and KNN method are chosen. Experiment result shows the SVM-KNN method is valid method,which has a higher classify racy than that of the SVM. Besides,the problem of choosing kernel function parameters SVM is solved using this method.