Using knowledge representation for agent world model

Abstract

This paper investigates the use of a description logic knowledge representation system, for representing and managing an agent's knowledge of the world for an intentional agent system. An existing application is represented using this mechanism, and the findings are discussed. A number of errors and/or inconsistencies were found in the original implementation using this methodology, indicating that it would be useful for consistency management. In addition there was a straight forward mapping from the domain knowledge to the knowledge representation system. An added benefit was a simplification of the plans required by the agent system.