A knowledge base for knowledge-based multiagent system construction

Abstract

The goal of the agentTool project at the Air Force Institute of Technology is to specify, design, and semiautomatically generate multiagent systems. The key to this ability is an underlying knowledge base that manages the knowledge used by the system designer in creating intelligent agent-based software applications. This paper describes the Agent knowledge Interchange Mechanism (AIM), the agentTool knowledge base. There are four main components to AIM: the knowledge parser, the Multiagent Markup Language, the AIM domain model, and the Agent-oriented Random-Access Meta-Structure (ARAMS). AIM is implemented using a multiagent system architecture that permits multiple developers to share a single knowledge base thus directly supporting collaborative system design and knowledge reuse.