Ontology Driven Concept Approximation

This paper investigates the concept approximation problem using ontology as an domain knowledge representation model and rough set theory. In [7] [8], we have presented a rough set based multi-layered learning framework for approximation of complex concepts assuming the existence of a simple concept hierarchy. The proposed methodology utilizes the ontology structure to learn compound concepts using the rough approximations of the primitive concepts as input attributes. In this paper we consider the extended model for knowledge representation where the concept hierarchies are embedded with additional knowledge in a form of relations or constrains among sub-concepts. We present an extended multi-layered learning scheme that can incorporate the additional knowledge and propose some classes of such relations that assure an improvement of the learning algorithm as well as a convenience of the knowledge modeling process. We illustrate the proposed method and present some results of experiment with data from sunspot recognition problem.