Using Ontology in Desinging a Multi-Agent System

Abstract

A multi-agent architecture is a distributed agents architecture which is derived from the distributed system concept and emphasizes on agent independence, and the cooperation between agents. For defining on intelligent agent, we divide a general agent framework into four components, the communication component, the control component, the agent knowledge component, and the execution monitoring component. A five-level approach has been proposed in agent system design: the detail level, the intermediate level, the policy level, abstract level, and the organization level. We apply the concept of ontology to implement the agent knowledge. The advantage of ontology is knowledge sharing and reuse, and these advantages can facilitate the cooperation between agents. The conceptual graph is a good approach to implement ontology. Using the technical of multi-agent system (MAS), we proposed an intelligent stock investment system as application.