Coupling Social Networks and Agent-Based Models: State of the Art and Prospects

Monday, 16 July 2018: 10:30
Oral Presentation
Meike WILL, Helmholtz Centre for Environmental Research - UFZ, Department of Ecological Modelling, Germany
Birgit MÜLLER, Helmholtz Centre for Environmental Research - UFZ, Department of Ecological Modelling, Germany
Jürgen GROENEVELD, Institute of Forest Growth and Forest Computer Sciences, TU Dresden, Tharandt, Germany
Agent-based modelling is a valuable tool for capturing the interactions of individuals or groups connected in a network. Therefore, it has been applied to different disciplines of contemporary research to model, for example, social-ecological systems concerning land use or resource management, epidemiology, or social dynamics. The arising topics like group formation, cooperation, or diffusion of information are similar throughout the different disciplines and the methods resemble one another. Although those methods are widely applied, reviews of the existing literature always focus either on one specific discipline or on certain interactions happening within the network; an overview across disciplines and methods is missing so far. To fill this gap, we systematically reviewed articles combining social networks and agent-based modelling. We divided the evaluation into categories concerning the underlying network structure and its properties, the characteristics of the agents, the links and interaction between them, and the incorporation of empirical evidence to the models. Based on this overview, we discuss to which extent an exchange across disciplines is necessary and useful, and where the commonly used practices are sufficient. While we identified a broad range of underlying network topologies describing the interacting populations across the different disciplines, ranging from data-based network structures to classical network structures like small-world or scale-free networks, we found that the interaction mechanisms show large varieties with respect to the context. For instance, models for information diffusion in the social sciences are much more detailed than those in epidemiology, where constant transmission rates for the diffusion of diseases are often assumed. Besides the aim to foster synergies between the disciplines, we focus on open challenges when dealing with social networks in agent-based models.