A Evolutionary Fuzzy ART Computation for the Document Clustering

Many clustering techniques have been widely developed in order to retrieve, filter, and categorize documents available in the database or even on the Web. The issue to appropriately organize and store the information in terms of documents clustering becomes very crucial for the purpose of knowledge discovery and management. In this work, a hybrid intelligent approach has been proposed to automate the clustering process based on the characteristics of each document represented by the fuzzy concept networks. Through the proposed approach, the useful knowledge can be clustered and then utilized effectively and efficiently. In literature, artificial neural network have been widely applied for the document-clustering applications. However, the number of documents is huge so that it is hard to find the most appropriate ANN parameters in order to get the most appropriate clustering  results. Traditionally, these parameters are adjusted manually by the way of trial and error so that it is time consuming and doesn’t guarantee an optimum result. Therefore, a hybrid approach incorporating an evolutionary computation (EC) approach and a Fuzzy Adaptive Resonance Theory (Fuzzy-ART) neural network has been proposed to adjust the Fuzzy-ART parameters automatically.