JULIA SEMANTICS LLC — Department of Defense SBIR Phase II: HR001121S0007-02

Description

As composite capabilities across warfighting elements multiply, modeling and simulation (M&S) analytics struggle to keep pace. Model integration complexity slows the development of responsive M&S for defense analyses. The challenge is that M&S mechanics are inaccessible, so modeling choices are hard to understand; M&S semantics—the meaning behind the mechanisms in M&S—require human interpretation, so complex integration tasks are labor-intensive; and modeling and integration choices, and policy, acquisition, and mission decisions based on M&S results are hard to defend, so the credibility of M&S-based analyses suffers. Today, semantics are discovered, curated, and interpreted by hand. M&S semantics are the why behind modeling and model integration choices and purpose to which the M&S community of users commits the M&S to practice. There are few, if any, computational methods to access semantics. Advances in the science underpinning M&S semantics are needed to realize computational methods and tools. Further, M&S creators perform complex M&S tasks, like integration, unguided by the theoretical principles necessary to support efficient, heuristic discovery, curation, and interpretation of M&S semantics. This means that the complexity of modeling, model integration, and integration testing scales nonlinearly but human M&S integration resources scale linearly. Current practice is limited by human capital. The opportunity is to apply predictive data analytics, machine learning and reasoning, and similar technologies to discover and curate M&S semantics so that simulation models can be reasoned about computationally—that is, computationally accessible semantics. Semantic Model, Annotation, and Reasoning Technologies (SMART) Semantic Analytics will develop prototype software tools for computationally accessible semantics. The SMART technology concept requires a solid foundation in simulation science, i.e., well-grounded in M&S formalism, to realize the computational methods needed to access M&S semantics in composable, constructive M&S architectures. With advanced M&S formalisms and theory enabling semantic composability fully formed, mature and commercially available semantic reasoning tools and technologies are readily available to create semantic analytics tools for model selection and other M&S integration tasks. Computationally accessible semantics will enable technologies that provide M&S creators with tools that augment complex M&S tasks like integration, improve insight to the performance of the integrated models, and buttress analytic conclusions drawn from M&S results—M&S consumers will benefit from faster results, reduced cost, and improved credibility of M&S-based analyses.