Observation and Analysis of Dynamic Networks in Agent-Based Models

Neufeld, Anna-Lena (2022) Observation and Analysis of Dynamic Networks in Agent-Based Models. Masters thesis, Institute for Visual and Analytic Computing, University of Rostock.

Text Anna-Lena_Neufeld_Master.pdf - Published Version Restricted to Registered users only Download (1MB)

Abstract

Agent-based modelling is a wide-spread technique to investigate the behaviour of a heterogeneous, interacting population. However, sociological studies often perform an analysis restricted to monitoring global properties of the resulting agent network. As most social networks exhibit a causal relation between the agents’ behaviour and the network structure, it makes sense to analyse the local properties and their impact on the entire population in more detail. The detection of key actors and lines of influence within the social network can provide insights in the network’s dynamics, as understanding its evolution in the past allows for predictions of its behaviour in the future. This thesis introduces a model checking approach to verify formal hypotheses on the simulation output data. The hypotheses concerning agent properties, agent constellations, network metrics and their combinations are formalised in linear temporal logic. The solution is applicable in any research context and was successfully tested on an existing agent-based simulation model.