The MASE design experience

Abstract

In Computer science, agent-based modeling (ABM) and multi-agent simulation contribute to scenario anal-ysis, since they may represent patterns of human interaction, which are essential to a better understanding of real systems, including many different aspects such as social, economical and environmental. This paper presents the use of agent-based techniques to define and develop the MASE - Multi-Agent System for Envi-ronmental simulation model using two different design approaches: the TROPOS Agent Oriented Software Engineering methodology with JADE framework, and the Agent-Object-Relationship modeling and simulation platform. MASE model system was defined to characterize land-use change dynamics through a replicable pa-rameterization process, useful to the development of agent-based simulation frameworks. However, different agent-based design experiences can enrich the modeling and development process of environmental simulation frameworks, since they embed conceptual structure definition according to specific design concepts. Thus, in this paper we include a short discussion of challenges, perspectives and limitations of two design experiences involving different concepts and techniques for information systems analysis. We believe the reported discus-sion can serve as a starting point to define evaluation criteria for comparative analysis of different agent-based design approaches. To illustrate land-use simulations with MASE we used the Brazilian Cerrado case study.