Using rough genetic and Kohonen’s neural network for conceptual cluster discovery in data mining

Abstract

We consider the problem of discovering the conceptual clusters from a large database. From Z. Pawlak’s information system in rough set theory, we define an information matrix, information mappings and some concepts in data mining literature such as large sets, association rules and conceptual cluster. We propose a combined method of information matrix, Kohonen’s neural network for large set discovery and genetic algorithm for conceptual cluster validity. We present an application of our method to a student database for discovering the rules contributing to the training of the gifted students.