MABS2000 @ ICMAS-2000

The Second Workshop on Multi Agent Based Simulation
8th-9th July, 2000

Background

The first MABS workshop, at the last ICMAS, had as its aim "to develop stronger links between those working in the social sciences, for whom agent based simulation has the potential to be a valuable research tool, and those involved with multi-agent simulation, for whom the social sciences can provide useful theories and exemplars." Since that workshop, discussions among social scientists and computer scientists in the multiagent systems community (for example in AgentLink, the European network of excellence for MAS research) have helped to focus the exploration of how representational social simulation and associated methodologies can inform the specification of agents and support the analysis of emergent properties of large-scale complex systems.

The second MABS workshop extends this development within a framework intended to provide for substantial discussion. The fifteen accepted papers fall broadly into three groups:

the lessons of social simulation for DAI more generally (Axtell, Schillo et al, David et al, Teran et al., Norling et al., Pedone-Conte)
papers supporting or reporting social simulation modelling (Sawyer, Hemelrijk, Rouchier et al., Downing et al.)
papers reporting social simulation based software applications (McGeary-Decker, Kafesa-Karlapalem, el Hadouaj et al, Breton et al, Davidsson).

The workshop attracted about 45 participants and was the largest of the federated ICMAS events. We succeeded in achieving informal and concentrated discussion of previously circulated papers. MABS2000 was hoped to be a watershed event in the development of agent based social simulation and the relationship of the social simulation community to the wider agents research community. I believe the participants would agree that this hope was realised. The overall theme of the workshop was the identification of key issues to be addressed by the agents community if representations of "messy systems" and applications for those systems are to be developed. "Messy" systems in this context are systems with unknown boundaries (if any) that are too large and intricate to be well understood by observers, modellers and developers. Such systems are not being addressed by the mainstream of the MAS community though there is some evidence of a growing recognition of the importance of doing so.

The proceedings will be published by Springer Verlag in the Lecture Notes in Artificial Intelligence series. This is a fully series.

A paper detailing the purpose of the workshop and how the various papers inform the development of agent based simulation is openly available here. Slides used to focus the final discussion of the workshop, updated to take account of points made in that discussion, are available from here in html and pdf formats.

The slides used by authors who have chosen to make them available will be linked from here when they have been provided.