Model-Driven Approach for Agent Based Software System

Starting from the implementation of software engineering, a model-driven development approach for agent-based intelligent applications is proposed in this paper on the basis of software architecture. Platform independent models for the agents are built firstly. Then, by integrating different functional agents such as user agent, management agent, task agent, the approach succeeds in transforming models at different levels. The target codes for the system are generated automatically. The experiment shows that this approach overcomes the limitations and the weaknesses of traditional agent system development, and improves the system quality and development efficiency.