Text clustering based on kernel KNN clustering algorithm

Document clustering is a popular tool for automatically organizing a large collection of documents. In this paper, we propose a Kernel-based K-Nearest Neighbor (KKNNC) clustering algorithm based on the KNN method. Our algorithm maps samples into a higher-dimensional feature space using a nonlinear function before clustering, then in kernel space divides them linearly. We also propose a new attribute selection method-ATS??algorithm, which can select important terms in documents. Our algorithm first uses ATS to eliminate redundant attributes in data sets, then gives each of the selective attributes a weight value according to the relationship between these attributes. The experimental results show that our algorithm is effective in the text clustering task. © 2013 by CESER Publications.