Developing social networks for artificial societies from survey data

Abstract

Authentically representing large social collectivities remains a preeminent challenge throughout the social computing, and modeling and simulation communities. We demonstrate here a simple technique that uses survey and polling data to embed agents with attributes and endogenously elicit an authentic and theory-driven simulation social structure for an artificial society. We furthermore show that a representation of social structure based on internal agent attributes allows for the continuous representation of social dynamics that affect agent cognition and association, and that social structures for artificial societies can be generated without any loss to the granularity of the underlying data or simulation output. We provide a case study using social survey data to demonstrate the method and effects, document the visualization of social structure for the population of Indonesia, discuss the implications and uses of survey data for social simulation, and suggest several paths forward for social and behavioral predictive modeling. © Springer-Verlag Berlin Heidelberg 2010.