Collaborative Research: Subnational Nonstate Actor Governance

The Sub-national Nonstate Actor Governance (SNAG) project introduces a new measurement strategy and public dataset to measure territorial control at the local level within conflict zones, tracked over time. Understanding how groups gain or lose territorial control, and thus how conflicts begin, evolve, and end, is essential to national security and preparedness.

Yet scholars, policymakers, and military strategists lack reliable and accessible techniques to measure and monitor territorial control within conflict zones. Existing empirical research is focused on a limited number of conflicts for which there happen to exist reliable measures of local-level territorial control over time. This limits ability to understand conflict more generally, and to apply knowledge to new threat environments.

This research draws upon open-source information to ensure a transparent process that is easily replicated across contexts and adapted to new measurement challenges. The project uses machine learning and natural language processing (NLP) tools to automatically detect mentions of belligerent activity and control in a corpus of open-source texts, which are then used to produce spatially and temporally disaggregated estimates of rebel and government territorial control.

The Subnational Nonstate Actor Governance (SNAG) project measures nonstate actors’ territorial control and governance at the local level, capturing temporal variation throughout conflict, comparable across contexts. This project makes both substantive and methodological contributions, generates new publicly available data capturing nonstate actors’ territorial control, uses an approach that translates across contexts to facilitate comparative analyses.

The PIs annotate text from a corpus of news reports from conflict zones, identifying indicators of rebel and government territorial control with location and time information. These annotations are then used to train a new natural language processing pipeline, which is applied to the remainder of the corpus to automate the process of extracting relevant information from the full corpus.

The information produced by this process is incorporated into a measurement model to produce fine-grained spatio-temporal data on conflict belligerents’ territorial control within conflict zones, facilitating systematic comparison of these phenomena within and across conflicts. The subnational territorial control data are used to investigate basic research questions related to the causes and consequences of territorial control and governance, fundamental to understanding the security risks in “differently governed” spaces, the efficacy of counterinsurgency aid, and the consequences for state-building after conflict.

Methodologically, SNAG contributes new tools for generating geospatial data from text and for developing spatial latent variable models adaptable for additional social science applications.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.