Machine Interface for Contracting Assistance (MICA)

ABSTRACT:To address the challenges posed by operating in accordance with complex federal acquisition regulations that discourage nontraditional providers from partnering with the government in emerging markets, ARiA LLC, a research-and-development firm that develops machine-learning and cognitive-computing software for commercial and government use, in collaboration with CPJ, a nonprofit nonpartisan civic-education and public-policy-research organization with unique expertise in translating government acquisition requirements for nontraditional providers, will develop the scientific and technical foundation for, develop the public-adoption strategy for, and demonstrate the feasibility of MICA, the Machine Interface for Contracting Assistance. MICA is a natural-language query system, reasoning engine, and recommendation system that leverages commercial-sector advances in cognitive-computing algorithms including IBM Watson cloud services, deep learning, and semantic web, to provide users from both government and current and prospective contractor organizations insight and clarification into what is required of them by defense-contracting statutes, regulations, practices, and policies. Beyond providing a clear, intuitive, and comprehensive natural-language query interface to defense-contracting statutes, regulations, practices, and policies, MICA leverages its semantic database of government requirements together with its advanced reasoning and machine-learning capabilities to automatically determine requirements for solicitations and provides those requirements to potential offerors in a clear natural-language format that supports drill-down.BENEFIT:The MICA cognitive-computing system to be developed and evaluated in this work will advance the state-of-the-art to significantly enhance the ability of small and innovative companies to partner with DoD in emerging areas of research. The Phase I effort will (1) develop a design architectural concept and early stage prototype for MICA, (2) demonstrate that MICA can feasibly be developed into a cognitive computing system that meets Air Force, DoD, and contractor needs, and (3) establish that MICA can be feasibly developed into a cognitive-computing application deployable on web-based platforms during Phase II and as a mobile app able to support all defense-acquisition-process phases and milestones during Phase III. This Phase III transition will be supported by a comprehensive rollout plan developed in coordination with CPJ, including marketing and training for prospective offerors and government officials. MICAs capability for automated generation of requirements associated with specific solicitations naturally extends to automated specification of required offeror capabilities, which further increases accessibility to nontraditional defense contractors and aligns with the continuing Department of Defense transformation to a capabilities based, net-centric acquisition approach. The natural-language query system, reasoning engine, and recommendation system we propose to develop and evaluate will significantly enhance the contracting process for new or nontraditional offerors across federal, state, and local government-contracting processes. It is aligned with open-government initiatives to enhance the acquisition process. Our solution is supported by large private foundations that seek to maximize return on their investment by aligning the work of their grantees with federal and state opportunities.