An Agent Scenario Mechanism Supporting Human/Agent Interaction

The Agent Oriented Programming (AOP) combines behaviors and states of software agent with code, so it would be somewhat difficult and complex for nonprogrammers who are knowledge experts such as salesman or social workers. In this research, we propose an FSDL which describes agent’s behaviors and reduces the difficulty and system complexity when a knowledge expert uses agent technique. FSDL supports a programmer a tool for writing a scenario representing knowledge expert’s experience and knowledge. It provides a tabledriven scenario editing interface for scenario designers to describe the desirable behaviors and states of software agents. The FSDL scenarios will then be transformed into the corresponding XML files which could be reusable for designing later agent scenarios. Finally, to demonstrate the feasibility of FSDL mechanism, we also set up an example system for promoting credit cards. Knowledge experts could therefore achieve the goal of driving a software agent to interact with human customers.