Fuzzy clustering for documents based on optimization of classifier using the genetic algorithm

Abstract

It is a problem that established document categorization method reflects the semantic relation inaccurately at feature expression of document. For the purpose of solving this problem, we propose a genetic algorithm and C-Means clustering algorithm for choosing an appropriate set of fuzzy clustering for classification problems of documents. The aim of the proposed method is to find a minimum set of fuzzy cluster that can correctly classify all training documents. The number of fuzzy pseudo-partition and the shapes of the fuzzy membership functions that we use the classification criteria are determined by the genetic algorithms. Then, the classifier decides using fuzzy c-means clustering algorithms for documents classification. A solution obtained by the genetic algorithm is a set of fuzzy clustering, and its fitness function is determined by fuzzy membership function. © Springer-Verlag Berlin Heidelberg 2005.