***** To join INSNA, visit http://www.insna.org *****
> I've had a little trouble finding a good introduction; these models
> seem to go by a wide variety of names. A couple hours of googling left
> me rather confused.
The main problem with multi-agent modeling is that 3 or 4 communities
that routinely use
its techniques hardly ever talk to each other, and speak very
different languages. Assumptions also vary widely, and greatly affect
the outcomes.
Here's a smattering of things to look for:
From Artificial Intelligence world - look for Belief/Desire/Intention
(BDI) model of agent behaviour;
agents tend to be complex but are embedded in simple social structures
(frequently all-to-all or star topologies); large numbers of agents
are infrequent. Also look for papers on SOAR.
Cellular Automata world - Wolfram, etc. - very simple agents in large
numbers, sitting on grids, reproducing complexity but not necessarily
modeling the exact phenomena.
Repast/NetLogo/Mason - in between the last two on agent complexity,
agents can be very simple or range up to some level of cognitive
accuracy; grids are fading away in favour of irregular neighborhoods
and networks
Multi-Agent Network Simulation - Construct, DyNet, NetWatch - agents
are cognitive automata, making locally optimal decisions based on
theoretical underpinnings (such as information diffusion theory).
Networks or proximities are not pre-assigned but a product of the
cognitive interactions and local rules.
I'm not giving you exact citations - but I think this should be enough
keywords to google for.
Max Tsvetovat
George Mason University
Center for Social Complexity
_____________________________________________________________________
SOCNET is a service of INSNA, the professional association for social
network researchers (http://www.insna.org). To unsubscribe, send
an email message to [log in to unmask] containing the line
UNSUBSCRIBE SOCNET in the body of the message.
|