Existing attempts to model civil conflicts, like those in Ukraine and Syria, are not currently reliable as they are often not data driven, or only rely on data provided by the states. The large volumes of open Internet and social media data available has the potential to rethink the way civil wars are modelled. How can we simulate and model civil conflicts in a data-driven way, to understand the dynamics of these events?
Explaining the science
Data-driven agent-based modelling is a powerful technique based on the design and implementation of interacting decision-making entities. Each entity assesses its environment and the state of the other entities and take a decision based on this assessment.
Using this technique, it is possible to study different “what-if” scenarios, and understand emergent behaviour given certain values of the system parameters. This technique has been applied in several areas, including economics, social policy, urban planning, amongst many others.
In this project, civil conflicts in geographic areas will be modelled in order to answer the following questions. Is it possible to model the dynamics of civil wars? What explains decisions to mobilise for a civil war?
Agent-based modelling, with the 'agents' representing different sides of a conflict, will be used to formulate dynamics between various groups. These include between central authorities and non-state armed groups, between different non-state actors, between adjacent states, and within individual armed groups.
Modelling the dynamics of civil wars is essential for improving our understanding of these events. This work has important implications for governmental and non-governmental organisations, for example in peace-keeping operations.