Agent-based modeling and simulation is critical to answering key questions within the Intelligence Community and Department of Defense. Ranging from strategic defense planning to public health and disaster recovery, many use cases involve emergent behavior where the actions of a few individuals can affect the behavior of the whole. Having actionable findings necessitates accurately depicting individuals’ motivations and decisions at scale. However,?modelers?often have limited access to open-source, accurate person-level data. To overcome this limitation and answer scenarios like these,?we?expand on known agent-based modeling capabilities to show geolocated, goal-directed behavior by synthetically representing a real population making real decisions.??
Prior work to generate synthetic?populations largely?utilized?aggregate?distributions commonly found in census datasets. Other work shows the importance of personality in shaping realistic behavior but has yet to generate representative individual agent personality. Our work builds upon methods in?the?behavioral?and computational sciences?by generating a population from available aggregate data for a specific geographic region to model personality effects on agents’ cognitive processes. Our goal is to better capture how the real-world population would respond to simulated scenarios.?We?detail how agents can be grown using representative aggregate data to guide that growth into realistic simulated entities.?We?generate agents using best-in-class methods within the Synthetic Reconstruction and Combinatorial Optimization classes.?Last,?we?discuss those entities taking motivated actions unique to their geographic areas.?
We?developed the current methods in collaboration with the National Geospatial-Intelligence Agency (NGA). Our work reveals the impact that population realism lends to applications across the Intelligence Community and Department of Defense. Implications for national security include higher?fidelity modeled?outcomes that inform?policy or?decisions when it comes to a populace, interventions, or making policies.?