QUARTERMASTER: Query and User-based Abductive Reporting Tool Enabling Responsive Multimodal Analysis of Simulated Technological Enterprise Records

At the Missile Defense Agency (MDA), the system-wide approach to test engineering generates massive amounts of multi-dimensional data, which analysts explore to evaluate Ballistic Missile Defense System (BMDS) configurations, engagement conditions, and target phenomena. Analyses are limited by time to construct each data query, technical knowledge, and size/complexity of the datasets and domain. To overcome these constraints, Aptima and Brandeis University propose to develop QUARTERMASTER (Query and User-based Abductive Reporting Tool Enabling Responsive Multimodal Analysis of Simulated Technological Enterprise Records). As an interactive cognitive assistant, it will leverage natural language processing, a domain ontology, causal inferences about users’ questions, and causal inference analyses of MDA simulation data to answer written and spoken queries by analysts. QUARTERMASTER will provide direct responses to queries and potential causes explicating the effect identified through abductive reasoning, and suggestions for clarifying simulation experiments should gaps remain in answering a query. QUARTERMASTER will augment MDA capabilities by enabling quick data sifting and understanding, and fast finding of key insights as a precursor to deeper, focused analyses. Ultimately, QUARTERMASTER will empower MDA analysts at all expertise levels to simply and naturally query their datasets, and thus contribute to faster and more complete analyses of BMDS. Approved for Public Release | 20-MDA-10398 (2 Mar 20)