A BDI Agent-Based Software Process.

Abstract

Agent-based programming comes us as a next generation programming paradigm. However, we have not been ready yet to fully use it without having sound and concrete software engineering methods and tools to facilitate agent-based software development. In this paper we propose a new software engineering process based on the BDI agent concept. We have refined and extend substantially our previous work, Agent-based Modeling Technique (AMT) and Agent-based Software Development Process (ASP), so that a systematic and realistic process has been born to construct BDI agent-based software. This paper introduces our new approach to the BDI agent-based software development process. The Belief-Desire-Intention (BDI) model has been proved as a dominant view in contemporary philosophy of human mind and action. We utilize BDI as a tool to analyze agents’ environments, goals, and behaviors. Use cases have been proved as a useful tool for requirement analysis. However, use cases cannot be neither agent-oriented nor object-oriented even though it has been used as a tool for analysis for a while. We have extended the existing use cases, and use a new sort of use cases to identify BDIs of agents in the real-world problem. We use external use cases to get the basic behaviors (intentions) needed to provide the services in the system. We use then internal use cases to define goals (desires) of the system and to discover more specified behaviors (intentions) to achieve the goals. By analyzing the behaviors (intentions) for each goal (desire), we can obtain environments (beliefs) on which the system behaves to perform the goal. The goal of this paper is to provide a very practical and systematic way to analyze and design the agent software based on the BDI concepts. We have started by using the existing proven tools and methods such as the use case approach, however, we have made a substantial modifications and improvements to these existing techniques so that we can analyze and design the system very realistically based on the BDI agent concept. This paper provides a systematic and seamless approach to the BDI agent-based software development. The way we suggest here to find BDI agents through requirements analysis is a unique and novel approach. This technique suggests a new way of thinking for BDI agent-based modeling.