Parts | Days | Selection | Search | Updates | Downloads | Help
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 22: Focus Session: Computational Social Science
SOE 22.5: Talk
Thursday, March 15, 2018, 17:15–17:30, MA 001
Next Generation Agent-Based Social Simulation — •Jan Ole Berndt and Ingo J. Timm — TriLabS@CIRT, Universität Trier, Germany
Agent-based social simulation (ABSS) has become a well-established research technique in computational social science (CSS). It complements network-theoretic and kinematics-inspired approaches to analysing emergent dynamics in complex systems. In addition, intelligent agents -- as researched from a distributed artificial intelligence (DAI) perspective -- provide information-processing, decision-making, and social interaction capabilities. Current ABSS is often either (a) limited in scale or (b) tends to oversimplify decision behaviour: (a) Cognitive models of motivations, emotions, and decisions adopted by DAI provide insights into comprehensible decision-making in small groups. (b) ABSS, e.g., with simple threshold models, enables analyses of large-scale networks. As CSS and DAI communities are rarely connected, this leads to a gap between simulations in DAI and social or cognitive sciences. We aim at bridging this gap to contribute to next generation social simulation. We discuss technical challenges of and methods for extending social simulation by sophisticated agent models, e.g., how to scale complex decision-making. Additionally, we address the problems of validating the resulting models and of using them to evaluate hypotheses in social simulations. We provide practical examples ranging from care demand forecasting and social media communication to social contagion of fertility.