Application of Model Driven Techniques for Agent-Based Simulation

Agent-based simulation is being recognized as a useful tool for the study of social systems. It is based on the idea that agents can be used as a good abstraction of members of a society, and by simulating their interactions, observe the emergent behavior. Usually, agent-based models for socialion of members of a society, and by simulating their interactions, observe the emergent behavior. Usually, agent-based models for social simulation are rather simple, but there are more and more works that try to apply this technique for rather complex systems, both in the typology of the agents and their relationships, and in scalability. For this reason, software engineering techniques are required that would facilitate the development of agent-based simulation systems. In this sense, it can be useful to consider agentoriented methodologies, as they cope with these requirements. This work explores the application of a model-driven methodology for the development of Multi-Agent Systems, INGENIAS. There are several advantages of this approach. One is the possibility to define specific modeling languages that are conceptually close to the domain-expert, in this case for simulation of social systems. This will facilitate the communication of multidisciplinary teams and, by generating code from models, the social scientist is alleviated from the concerns of programming for a simulation platform. This is illustrated with a case study of urban dynamics.