|
Activity Number:
|
18
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statisticians in Defense and National Security
|
| Abstract - #306090 |
|
Title:
|
Agent-Based Methods for Dynamic Social Networks
|
|
Author(s):
|
Eric Vance*+ and David Banks
|
|
Companies:
|
Duke University and Duke University
|
|
Address:
|
114 Old Chemistry, ISDS, Durham, NC, 27705,
|
|
Keywords:
|
agent-based models ; social networks ; social space
|
|
Abstract:
|
Agent-based models of social network interactions allow simple encoding of complex social expectations and conventions. For static networks, traditional mathematical formulations (e.g., the p1 model by Holland and Leinhardt (1981) and its generalizations) can be tuned to model almost any multiway relationship, using the clique-decomposition of Frank and Strauss (1986). But this is cumbersome and the formalism does not extend in natural ways to time-varying networks. The notion of actors in a social network occupying positions in a latent social space was introduced by Hoff, Raftery, and Handcock (2002). Our approach uses agent-based models with rules constructed to correspond to terms in a p1 model and rules determining how agents move dynamically in social space. Our talk will illustrate attractive features of agent-based models in such situations.
|
