Integrating Different Knowledge Representations in an Intelligent System: Standardization Allows Diversity?

This paper proposes that for a complex intelligent system, the choice of a knowledge representation language for the interaction processing component and the domain reasoning component should be left unconstrained provided that some type of common knowledge representation language for communicating oals, actions, and inputs is available. This common language would be used for both components for information exchange. In this way, it may be possible to design a domain-independent interaction processing component that is applicable to a genre of interactions such as advisory, database query, or task-assistance interactions.