Application of layered clustering and plane partition in Web page classification

The layered clustering can create layered nesting class with high precision. But the computing complexity is relatively high so that it is not fitted to solve large amount of sample calculation problems. K-means method has high efficiency while it is affected easily by the choice of the center of initial clustering. So to the irregular distributed samples, the clustering effect is usually not good. The paper focuses on the distribution features and complexities of samples in Web page classification and puts forward a classification method - to combine the layered clustering and plane partition. Firstly we use the algorithm of layered clustering in a few of samples to generate original clustering centers of sample set. Secondly, K-means method is used to classify the whole samples set. This strategy not only takes full advantage of the high efficiency of the K-means algorithm but also makes good use of the high precision and reliability of layered clustering method. Finally, this paper apply the method of combining the layered clustering and plane partition to solving the problems of text classification and presents some experimental results.