B. Ganter and G.W. Mineau (Eds.): ICCS 2000, LNAI 1867, pp. 113-126, 2000.  Springer-Verlag Berlin Heidelberg 2000 Discovery of Class Relations in Exception Structured Knowledge Bases Hendra Suryanto and Paul Compton Department of Artificial Intelligence, School of Computer Science and Engineering, University of New South Wales, Sydney, Australia {hendras,compton}@cse.unsw.edu.au Abstract. Knowledge-based systems (KBS) are not necessarily based on a well-defined ontologies. In particular it is possible to build very successful KBS for classification problems, but where the classes or conclusions are entered by experts as free-text sentences with little constraint on textual consistency and little systematic organisation of the conclusions. This paper investigates how relations between such ‘classes’ may be discovered from existing knowledge bases. We have based our approach on KBS built with Ripple Down Rules (RDR). RDR is a knowledge acquisition and knowledge maintenance method which allows KBS to be built very rapidly and simply by correcting errors, but does not require a strong ontology. Our experimental results are based on a large real-world medical RDR KBS. The motivation for our work is to allow an ontology in a KBS to ‘emerge’ during development, rather than requiring the ontology to be established prior to the development of the KBS. It follows earlier work on using Formal Concept Analysis (FCA) to discover ontologies in RDR KBS. 1 Introduction Most knowledge acquisition methodologies first build a model of domain knowledge before applying this to building a particular problem solver e.g KADS and CommonKADS [20], Protege2000 [24]. Although this approach facilitates re-use it does not overcome the knowledge-acquisition and maintenance bottleneck and these problems are present both in the development of the ontology and consequent problem solver. The RDR approach starts knowledge acquisition (KA) to build the problem solver immediately without any modelling apart from a simple attribute-value data representation [17]. Even the attribute-value representation can be developed while KA is in progress. The focus of the approach is to make the addition of each incrementally added piece of knowledge as simple and as reliable as possible. Although this approach facilitates KA and maintenance, it does not facilitate re-use because of the lack of an ontology. Richards and Compton [18] have previously applied formal concept analysis (FCA) to develop conceptual models from RDR systems and for example have discovered interesting models from a 60 rule blood-gas KB which was itself part of a larger RDR KB. The RDR approach, along with other KBS approaches, allows a