A contribution to modification of PART clustering algorithm for text processing

The paper presents one of the possible modifications of the Projective Adaptive Resonance Theory (PART) clustering algorithm and its application in the processing of text documents. Clustering on the basis of PART can be applied, for example, to generate a dictionary of keywords from a text. The principle of this method is based on clustering of words with the same root word. In order to demonstrate the increased clustering accuracy, the achieved clustering results with modified algorithm are compared with the results based on the standard PART algorithm.