Applied Knowledge Representation

The number of knowledge representation (KR) formalisms grows exponentially. What most of those formalisms lack is a successful application. This is particularly true of logic-based KR systems, often created without taking tractability into account. Applied Artificial Intelligence (AI) systems, on the other hand, are often based on one specific formalism claimed to be the one. In this paper we present several knowledge representation tools used within one project and show relationships between these languages. In this way we try to support the conjecture that when designing a complex AI system, one should use various KR formalisms, namely the ones most suitable for purposes of different subsystems, and devise an appropriate tool for data (or knowledge) interchange between these subsystems. The KR formalisms need not be the ones traditionally associated with AI — there are many other valuable tools. We are using one particular example of an AI system being designed within the PROMETHEUS project.