Simulating Agent Societies with Presage

  

Brendan Neville

 

Bio:
Brendan Neville is a Research Assistant in the Intelligent Systems and Networks Group at Imperial College London. He received his Masters from Imperial College London in 2002. Since graduation he has worked on several European funded research projects including Alis (Legal reasoning, Game Theory, Computational Logic), Tiramisu (DRM) and Alfebiite (Agent Societies). His research interests include Simulation of Agent Societies, Self-organising social-networks, Trust and Reputation. He is currently completing his PhD on the simulation of agent societies; the simulation platform Presage used in the thesis will soon be available as open source from http://sites.google.com/site/presageproject/

Abstract:
Agent societies provide a compelling computational paradigm for building applications whose chief characteristics is that they are open, volatile and decentralised.

In this talk we will present the Presage platform which can be used for simulation, animation and rapid prototyping of agent societies, facilitating investigations into the complex social behaviour of components, the evolution of network structures, and the adaptation of conventional rules.

We will describe a range of experiments with agent societies that use Presage, including a socio-cognitive model of trust, risk and reputation for agent mediated e-commerce; a social recommendation system for peer to peer networks; an intellectual property game for identifying, avoiding and resolving legal disputes; and the organised adaptation of institutional rules for ‘fair’ resource allocation in decentralised systems.

The talk will conclude with a discussion of future opportunities including experiments with heterogeneous agent architectures, games and simulations involving humans and agents, time prioritization vs. planning and the role of trust as a cognitive shortcut.