© Copyright JASSS

  JASSS logo ----

François Bousquet, Paul Davidsson and Jaime Sichman (2003)

Report on the Multi-Agent Based Simulation (MABS) 2002 workshop, Bologna, Italy, July 2002

Journal of Artificial Societies and Social Simulation vol. 6, no. 2

To cite articles published in the Journal of Artificial Societies and Social Simulation, please reference the above information and include paragraph numbers if necessary

Received: 18-Mar-2003      Published: 31-Mar-2003

* Abstract

Multi-Agent Based Simulation (MABS) workshop is a key event for the researchers working in the field of agent based simulations because it favours the encounter between researchers woking on applied social simulations who are interested on computer science aspects (architectures, platforms, methodologies) on the one hand, and computer scientists willing to understand how agent based models can be modelled from the observation of the real world.

Interdisciplinarity; Methodologies

* Introduction

MABS 2002 was organized in Bologna, Italy, in July 2002, as a workshop of the conference AAMAS (Autonomous Agents and Multi-Agent Systems). This was the third workshop following the ones organized in Paris (1998) and Boston (2000). Revised versions of the papers presented at these workshops were published as volumes 1534 and 1979 in the Lecture Notes on Artificial Intelligence series. The aim of MABS workshops is to develop stronger links between researchers working in social sciences and researchers involved with multi-agent simulations and computer science. The accepted papers were revised and published as an issue of Lecture Notes in Artificial Intelligence (Sichman, J., Bousquet, F. and Davidsson, P., editors. 2003. Multi-agent-based simulation II. LNAI 2581, Springer).

We are pleased to note that many important conferences in various disciplines such as geography, economics, ecology, sociology, and physics have hosted workshops on MABS-related topics and that many respected journals publish papers that includes elements of MABS. But although MABS is gradually acquiring legitimacy in many disciplinary fields, much remains to be done to clarify the potential use of MABS in these disciplines. Researchers from these disciplines have different points of view on issues such as time-frame, space, geographical scales, organisational levels, etc.. Moreover, the interest for MABS goes beyond the scientific communities. Several MABS models has been developed and used interactively with other societies. For instance, research is being done on the interactions between societies of robots and groups of people, and simulation models are developed with stakeholders for environmental issues in a participatory way, through the Internet or directly on the field. These new approaches lead to new research questions regarding the use of MABS for collective decision making, but also regarding the conceptual and technical aspects of MABS.

Within this framework of interactions between artificial and human societies, special attention was given to the conceptual and technical aspects (agent architecture, interaction protocols, simulation platforms, modelling protocols, time and space representation, presentation of simulation results) resulting from these interactions and favouring them.

Twenty-six papers were submitted to the workshop. After having been reviewed by at least by two referees, 12 were accepted for presentation. The selected papers were grouped in three categories:
  1. papers on emergence, alliances and groups,
  2. papers on MABS platforms and languages,
  3. papers on MABS applications.

At the workshop Alexis Drogoul, from Paris 6 University, gave an invited talk and an article by him and his colleagues has been added to those accepted for the workshop. This invited paper gives a fair idea of the present discussions and trends in the MABS community which involves "the thematicians, the modellers and the computer scientists". Bruce Edmonds, in his call for an ideal social simulation language also considers these two roles, the computer scientist and his tool and the modeller and his theories. The discussion in Bologna and the papers in the LNAI volume shows the confrontation between the thematicians and modellers in the one hand and the computer scientists in the other hand.

From the point of view of the computer scientist, Drogoul states that "MABS, despite its name, is in fact rarely based on computational agents... the notion of agent although shared at a conceptual level by different participants, does not imply the systematic use of computational agents in the system deployed". This point of view is supported by the set of papers on platforms and languages, based on computer scientists' issues and concepts. A group of researchers from Portugal and Brazil (David and co-authors, Marietto and co-authors) present two papers on requirements of MABS platforms and software process for agent based simulation. The paper of Jonker & Treur falls in the same category: they propose a conceptual framework to understand how organizational structure relates to dynamics. Amiguet proposes a new platform which focus on the dynamics of social networks. For the two last papers the Agent Group Role model is discussed and implemented.

The position of the computer scientist is challenged by the thematicians and modellers as stated in the Jager & Janssen paper which calls for "agents rules based on theoretically rooted structure that captures basic behavioural processes. Such a structure should be based on state-of-the art behavioural theories and validates on the micro-level using experimental or field data of individual behaviour".

When examining the papers with this classification in mind (agent based on computer science concepts vs. agents based on behavioural theories), one will find:
  • Two papers using agent models based on social theory. The paper of Jager & Janssen which base their models on the consumat approach, and the paper of Winoto based on an economic theory of crime. The paper of Hales who uses tag models may belong to this category although it is debatable.
  • Four papers using agent architectures conceptualised by computer scientists. Antunes & co-authors propose a belief-values-goals architecture to implement Axelrod's tribute model, Amigoni & Gatti propose an architecture called anthropic agency for physiological processes, and Machado & co-authors test different architectures to perform a patrolling task.

In the field of multi-agent based simulation there are still different posture, depending mainly on the scientific community of the researcher. A modeller or a computer scientist, working on the same issues of agent based simulation will orient their research in the same directions. A good example is the comparison of the papers of Marietto & co-authors on the one hand and Edmonds on the other hand who both define the needs for future platforms, but from different points of views.

Now that several workshops on agent based simulations in different scientific fields exist, MABS remains the very place where computer scientists and researchers from other disciplines interact and exchange ideas. To strengthen these interactions, MABS III which will be organized in 2004 at Melbourne, will request that the authors explicitly express the implication of their research for the complementary disciplines.

* Papers presented at the workshop

Multi-Agent Based Simulation: Where Are the Agents?
Alexis Drogoul, Diane Vanbergue and Thomas Meurisse (University Paris VI, France)

Emergence, Alliances and Groups
BVG Choice in Axelrod's Tribute Model
Luis Antunes (University of Lisbon, Portugal), Leonel Nobrega (University of Madeira, Portugal) and Helder Coelho (University of Lisbon, Portugal)

Evolving Specialisation, Altruism and Group-Level Optimisation Using Tags
David Hales (Manchester Metropolitan University, UK)

The Need for and Development of Behaviourally Realistic Agents
Wander Jager (University of Groningen, The Netherlands) and Marco Janssen (Indiana University, USA)

Relating Structure and Dynamics in Organisation Models
Catholijn M. Jonker and Jan Treur (Vrije Universiteit, The Netherlands)

MABS Platforms and Languages

The MOCA Platform: Simulating the Dynamics of Social Networks Matthieu Amiguet (University of Neuchâtel, Switzerland), Jean-Pierre Muller (CIRAD, France), José Baez and Nagy Adina (University of Neuchâtel, Switzerland)

Towards an Emergence-Driven Software Process for Agent-Based Simulation
Nuno David (ISCTE, Portugal), Jaime Simao Sichman (University of Sao Paulo, Brazil) and Helder Coelho (University of Lisbon, Portugal)

Towards an Ideal Social Simulation Language
Bruce Edmonds (Manchester Metropolitan University, UK)

Requirements Analysis of Agent-Based Simulaton Platforms: State of the Art and New Prospects
Maria Bruno Marietto (University of Sao Paulo, Brazil), Nuno David (ISCTE, Portugal), Jaime Simao Sichman (University of Sao Paulo, Brazil) and Helder Coelho (University of Lisbon, Portugal)

MABS Applications
On the Simulation of Multiagent-Based Regulators for Physiological Processes
Francesco Amigoni and Nicola Gatti (Politecnico di Milano, Italy)

Multi-Agent Patrolling: An Empirical Analysis of Alternative Architectures
Aydano Machado, Geber Ramalho (Federal University of Pernambuco, Brazil), Jean-Daniel Zucker and Alexis Drogoul (University Paris VI, France)

On Multi Agent Based Simulation of Software Development Processes
Tham Wickenberg and Paul Davidsson (Blekinge Institute of Technology, Sweden)

A Simulation of the Market for Offenses in Multiagent Systems: Is Zero Crime Rates Attainable?
Pinata Winoto (University of Saskatchewan, Canada)


ButtonReturn to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, [2003]