SBIR-STTR Award

A central challenge for the U.S. armed forces is trying to forecast how host nation populations may respond to both planned and unplanned military activities. These activities include natural and human caused events that may be kinetic or non-kinetic in nature. To address this issue, we propose to develop the Forecasting Activities through Cultural, Epidemiological, and Temporal Semantics (FACETS) system. FACETS will allow company level and above military planners and leaders to use agent-based models of a given population to forecast likely opinion changes of individuals and groups to certain events. FACETS will be rooted in theories of social identity and influence to better shape the socio-cultural landscape of a given population. Furthermore, FACETS will apply epidemiological inspired models of information spread to help forecast when and where opinions to events will change within a population. To accomplish this, FACETS will incorporate latent semantic variable techniques to find relevant events, topics, and associated opinions within a wide span of internet data sources which will serve as the informational and event space of the population.

Keywords:
Cultural Modeling, Opinion Forecasting, Identity Groups, Latent Variable Analysis, Natural Language Processing, Course Of Action Development, Epidemiological Modeling

A central challenge for U.S. forces is anticipating how foreign populations may respond to activities or incidents (e.g., military actions, natural disaster, psychological operations campaigns). Unfortunately, natural changes of attitudes and opinions within a particular population complicate our efforts to assess their opinions and possible reactions to different courses of action. To address this difficulty, Aptima proposes to develop FACETS (Forecasting Activities through Cultural, Epidemiological, and Temporal Semantics). FACETS will provide military planners the capability to forecast shifts in opinion over time within a population of interest. These types of tools have traditionally relied on surveys and censuses, which are accurate but costly, time-consuming, and time sensitive. To compensate for these shortcomings, Aptima will apply natural language processing techniques to extract relevant events, topics, and opinions from population-authored data (e.g., internet news, blogs, and social media). These data will then be used to simulate reactions to current events within the agent-based model, thereby extending the useful life of survey-based analyses. FACETS will then incorporate the mathematics of epidemiology to determine how information spreads within the population. Aptima’s approach takes advantage of often under-utilized data sources to more accurately and efficiently determine whether we are winning the battle of “hearts and minds.”

Keywords:
Sociocultural Modeling, Opinion Forecasting, Spatiotemporal Modeling, Agent-Based Models, Latent Variable Analysis, Natural Language Processing, Course Of Action Development,