© Copyright JASSS

  JASSS logo ----

Rosaria Conte and Scott Moss (1999)

Special Interest Group on Agent-Based Social Simulation

Journal of Artificial Societies and Social Simulation vol. 2, no. 1, <http://jasss.soc.surrey.ac.uk/2/1/4.html>

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-Jan-99            Published: 31-Jan-99


Keywords:
Agent-based Social simulation, Special Interest Group, AgentLink

The Agent Based Social Simulation (ABSS) Special Interest Group (SIG) has been formally recognized by the AgentLink management committee. AgentLink is Europe's network of excellence for agent-based computing, funded by the ESPRIT programme. As with all AgentLink SIGs, it has a budget to fund attendance at SIG meetings and to support the development of collaborative European projects.

Computer simulation has proved useful for modelling phenomena of traditionally social scientific interest, such as cooperation, coordination, organizational behavior, social dynamics, group and coalition formation, and the evolution of conventions and norms. Multi-agent researchers soon came to realize how crucial these topics are within their field. In particular, the study of emergence of social phenomena such as organizational performance and optimization, cultural norms, institutional forms has become a major direction of research in MAS. In turn, such social modelling brings into play a variety of normative concepts, such as conventions and obligations, and phenomena, such as commitment and responsibility, and draws attention to how these phenomena evolve among computational agents in interaction. These concerns have led social simulators to pay increasing attention to agent modelling. Dissatisfied with the model of the rational social actor, they have developed simulation models of evolutionary social phenomena incorporating representations of cognition derived from other disciplines such as cognitive science and social psychology. However, the model of the agent used is essentially behavioural and frequently more rudimentary than those developed in some areas of AI.

The computational study of social organizations and institutions is a topic of growing interest in both the computer science and social science communities. In the formal/computational scientific communities, logical philosophy and social philosophy have a long tradition in studying institutions and obligations. Interest in such issues is rapidly growing. This is shown by several indicators, including

  1. the number of workshops, etc.;
  2. the diffusion of notions of obligation, convention, trust, commitment, reciprocity, right, permission, etc. in the research on intelligent agents; and
  3. the attention paid to norm-based phenomena in designing and implementing situated intelligent agents (consider the trade-off between robust performances and flexibility: formal and computational research on commitment essentially proceeds from the question of how to design intelligent, adaptive, flexible agents that exhibit robust performances; moreover, think of the research on conventions as solutions to problems of coordination among autonomously interacting agents).

Interestingly, the more the MAS researchers pay attention to evolutionary and dynamic organizations and institutions, the more they use computer simulation (for example, simulation-based studies on the evolution of conventions, commitment, altruism, in MAS, and the role of simulation in the study of organizations). Where the social sciences meet the physical and biological sciences as, for example, in the modelling of climate change, there is growing disenchantment with analytic equilibrium approaches to analysis of social and economic issues. Agent-based simulation methods are proving attractive to physical scientists seeking to take socioeconomic factors into account in the analysis of such issues. Increasingly, we are seeing agent-based social simulation used to provide more realistic alternatives to analyses of the whole area of exchange with particular success in generating empirically more satisfactory models of financial markets.

For these reasons, the social simulation research community has developed rapidly in recent years, attracting sociologists, computer scientists, logicians and economists (and this list is not exhaustive) who believe that simulation methods can support the analysis of social issues while maintaining the relevance of sociology and the rigour of economics. The result is a happy research community in which participants from different discipline backgrounds use one anothers' techniques and approaches to develop new methods and methodologies. We have seen the use of formal logics to represent such concerns of sociologists as trust, belief and helpfulness. Results and representations from cognitive science are regularly used to specify agents in computational organization theory and a new approach to economic issues unfettered by old, conventional approaches.

This SIG is intended to reinforce the development of a strong European presence in social simulation and also to demonstrate that the disciplines on which the social simulation community draws can themselves benefit from the results obtained by members of that community. Perhaps because of its relative youth or because it encompasses an exceptionally diverse set of academic backgrounds, the social simulation community remains open to new ideas and to new applications. This is a strength on which the SIG is intended to build.

ABSS is not intended to be another home for conventional economics. Applications of equilibrium representations of markets and game theoretic approaches to, for example, electronic commerce are already widespread and catered for by other AgentLink SIGs.

The first ABSS workshop will be co-located with the PAAM conference in London on the 21st and 22nd April, 1999. All of the various social simulation approaches will be represented at that workshop. In keeping with the AgentLink approach, this will not be a meeting to present and listen to academic papers about completed work. The workshop is forward looking, participants are expected to present speculative papers about their work and how it might benefit from and support future developments in agent-based social simulation. The workshop will therefore involve academics and commercial users in discussions about the agenda for agent-based social simulation research in Europe.

The programme has been devised to support exchanges among users and applied social simulation modellers, those who develop conceptual frameworks for agent-based simulation from, for example, the deontic logic community, developers of software environments for modelling such as SDML, CORMAS and MASSIF. Applications projects reported will include model-based analysis of group formation, behaviour and disintegration, integrated assessment of climate change and urban issues.

----

ButtonReturn to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, 1998