M2M2

Second “Model to Model” workshop:

Comparing and combining agent-based models with each other,
and with other models
Workshop to be held on 16-19 September 2004, Valladolid, Spain

In association with the second ESSA conference

• Aims and Topic

• Date and Venue: Change of Date

• Call for Papers

• Organising committee

• Local organisation

• Program committee

• Provisional program

Aims and Topic

 In recent years there has been an explosion of published literature utilising Multi-Agent-Based Simulation (MABS) to study social, biological and artificial systems. However, despite the plethora of novel models and interesting results it is rare that models are compared, built-on or transferred between researchers. It would seem there is a dearth of  "model-to-model" analysis.

The second M2M workshop, M2M2, is aimed at gathering researchers in MABS who are interested in understanding and furthering the transferability of knowledge between Multi-Agent Based Simulation models and beyond. As in the first M2M workshop (Marseille, France, March-April 2003), papers are invited reporting comparisons and validations involving two or more models, at least one being an agent-based model.

Understanding complex systems often seems to necessitate the use of more than one model.  By specifically comparing models a better view of what modelling brings to the understanding of (real and artificial) societies may be facilitated. The union of two or more models with apparently incompatible levels of abstraction, comparisons of results from related studies, and the description of model findings in ways allowing others to check and replicate them, all require the development of methods that improve rigor and reproducibility.

Approaches of interest include but are not limited to:

• Rewriting models that others have described in papers so as to understand them more deeply and reproduce the stated results;
• Composition of models where different scales are inter-related in a larger model – the results of one model being used in the other;
• Aligning of models: a comparison of different models that announce the same type of results, so as to evaluate their actual similarities;
• Comparison of different models based on their fitness to a set of data, accumulated through field studies or experiments;
• Using one model as a post-hoc summary or abstraction of another model's results; Constraining the scope of an existing model to enable more powerful techniques to be applied in a different computational framework;
• Using models with different structures and assumptions to confirm each other's results;
• Determining what to do when two models give results that contradict each other.
• Constructing and using taxonomies or description schemas for use in comparing or combining MABS models.

At the first M2M workshop, two specific issues were addressed:

(1) How can different multi-agent models best be compared? (What kinds of comparison are feasible, and how can comparisons be useful?)

(2) Is it possible to transmit the structure of a model and its results among researchers through papers, or should we look for other means of transmission?

The meeting proved that the issues at stake seemed fundamental for many practitioners. Most papers addressed the replication of models through alignment: comparing different level of aggregation describing the same system and evaluating the differences and similarities between centralised models and distributed ones; checking the influence of rationality at different result levels, or trying to integrate several forms of rationality. Other papers addressed the issue of replication and concluded here that there are considerable conceptual and practical difficulties in replicating others’ work, but doing so can be most enlightening.

A number of papers have been published in a resulting special issue in JASSS[1]. Authors wishing to submit a paper to M2M2 are advised to refer to this issue, although it is not representative of all possible approaches on the topic. 

Following M2M, this new workshop will hopefully bring together an international community and produce new frameworks to look at agent-based simulation results.

Date and Venue: Change of Date

The M2M2 Workshop was scheduled to be held on 14 and 15 September 2004, just before the second ESSA (European Social Simulation Association) conference.  It is now intended to hold the workshop as parallel sessions within the ESSA conference, which runs from 16-19 September, cutting accomodation costs for attendees, and making it unnnecessary for a separate attendance fee to be charged. We would nonetheless like to apologise for this change of plan. Please send any queries to Nick Gotts (n.gotts@macaulay.ac.uk).

Organising Committee:

Claudio Cioffi-Revilla

Guillaume Deffuant

Nick Gotts

Juliette Rouchier

Call for Papers

Send paper by email by April 15, to any member of the Organising Committee. Papers should be sized between 10 and 20 pages and will be selected through peer review. Contributors wishing to submit demos, models, or other material with their paper are invited to contact a member of the organising committee.

Organising committee

Claudio Cioffi-Revilla ccioffi@gmu.edu

Guillaume Deffuant guillaume.deffuant@cemagref.fr

Nick Gotts n.gotts@macaulay.ac.uk

Juliette Rouchier rouchier@ehess.univ-mrs.fr

Local organisation

 Cesáreo Hernández Iglesias essa@eis.uva.es

Program committee (to be confirmed)

Rob Axtell, Francois Bousquet, Paul Davidsson, Marco Janssen, Christophe Le Page,
Scott Moss, Mario Paolucci, Javier Pajares, Adolfo Lopez, David Hales, Bruce
Edmonds,  Frederic Amblard, Bill McKelvey, Luis Antunes, Olivier Barreteau, Matt
Hare, Thierry Faure, Thomas Brenner, Wander Jager, Catholijn Jonker.