OUSD (R&E) MODERNIZATION PRIORITY: Artificial Intelligence/Machine Learning TECHNOLOGY AREA(S): Information Systems The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the statement of work (SOW) tasks intended for accomplishment by the FN(s) in accordance with the Announcement. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws. Please direct questions to the Air Force SBIR/STTR HelpDesk: usaf.team@afsbirsttr.us. OBJECTIVE: Develop knowledge graph based analytical software to enable knowledge acquisition at finger tips and meaning making at scale on red force behavior and operation dynamics from all-source intelligence data to support ISR operation planning and management. DESCRIPTION: Through the entire Joint Air Tasking Cycle (JATC), many ISR planning and analytical tasks require a good understanding of red force behavior and operation dynamics. For example, in order for ISR analysts to translate the commander's intent into a clear, concise, accurate, and relevant set of collection requirements (CRs), they have to acquire and constantly update the knowledge on an adversary's conducts, states, and intentions from all-source intelligence. However, the helpful information is often buried in huge volumes of disparate and uncorrelated raw intelligence data without apparent answers to these questions. This makes the current time-bound CR development a cognitively intensive manual process. It is difficult to scale it up into a high-intensity near-peer operational environment where the hidden dynamics of a large red force operation are too complex for any individual analysts to mentally digest and remember in real time. Therefore, there is a critical need in the area of integrated ISR by the Air Combat Command (ACC) for new machine-assisted knowledge acquisition and meaning making capability to augment analysts for continuous acquiring, retaining, analyzing, understanding, and forecasting of red force behavior and operation dynamics from massive and noisy real-time as well as historical all-source intelligence data. The advancement in the artificial intelligence has offered some potential solutions to address the problem, particularly in the domain of knowledge graph (KG) which has witnessed large commercial success in Google search and Amazon's Alexa for providing comprehensive search returns on individual query targets as well as their correlated entities. In this effort, AFRL is seeking innovative solutions on KG model and additional machine inference of red force operational behavior and dynamics so analysts can have relevant red force information at finger tips and mean-making at scale when working on analytical JATC tasks. The definition of KG is broad in this effort and not limited to specific modeling technology such as the traditional ontology-based models. Any connectivity-focused, analytical solutions are highly encouraged. More specifically, AFRL looks for a software solution that can deliver a scalable KG design and corresponding graph database, data processing modules, data analytical engine, and front-end graphical user interface (GUI) and visualization. It should be capable of modeling, detecting, forecasting, and visualizing red force operational tactics, techniques, and procedures (TTP) in the form of spatial-temporal operational patterns of units and weapon systems, indicators of state changes, and group interactions at tactical and joint operational levels. The KG design should include relevant combat, support, and command and control components with group behavior and risk models in order to derive information on red force's posture, intent, operation mode, and psychological state. It also needs to be flexible on architecture and fault-tolerating with respect to missing or uncertain intelligence data. The analytical engine should provide confidence levels in its analytical results and summary statistics to facilitate sound decision making process. The data processing modules need to be able to extract and parse spatial-temporal information from multiple representative intelligence sources, including open sources. The GUI should allow analysts to easily construct query and provide user-friendly presentation of analytical results in the form of annotated graphs, maps, tables, and/or charts, etc. The operational scenarios may include, but not limited to, ground to air and air to air engagements. AFRL will provide a limited number of simulated datasets for phase I and II. The use of government datasets is optional as long as the offeror's own datasets are clearly identified in the proposal. Open source datasets are highly encouraged. No other government furnished materials, equipment, data, or facilities will be provided. PHASE I: Design and develop the initial software architecture and critical components for a proof-of-concept demonstration involving a few tactical level scenarios using simulated and open source data. The focus is on graph model and backend analytic engine. Provide trade-off analysis on the best technical development path, algorithm and method choice, data management and software framework decision, and potential risk and negation strategy. PHASE II: Develop all aspects of a fully functional prototype with a user-friendly front interface and scalable backend data process and management. It needs to deliver a seamless modeling and analysis pipeline at both tactical and joint operation levels of the red forces using simulated, multi-domain open source, and other DoD internal data. Conduct test and validation with AFRL and ACC analysts to demonstrate the human performance difference against current practice for a specific JATC task, for example, development of PIRs (Priority Intelligence Requirements) or EEIs (Essential Elements of Information) and/or facilitating asset management and task assignment. PHASE III DUAL USE APPLICATIONS: Adapt, refine, and optimize the Phase II prototype into a mature product directly integrated with analytical systems at one of ACC's Air Operation Centers to support multiple JATC tasks, for example, CR development, asset/task pairing, and battle damage assessment, using real mission data. Expand the software into other DoD branches such as the Space Force as well as the commercial world for applications in disaster relief [1], law enforcement [2], and many other areas [3]. REFERENCES: Gaur, M., Shekarpour, S., Gyrard, A. and Sheth, A., empathi: an ontology for emergency managing and planning about hazard crisis, Proc. IEEE 13th International Conference on Semantic Computing (ICSC), pp. 396-403, (2019). Kejriwal, M., Szekely, P. and Knoblock, C., Investigative knowledge discovery for combating illicit activities, IEEE Intelligent Systems, 33(1), pp.53-63 (2018). https://neo4j.com/use-cases/ KEYWORDS: Multi-Domain Command and Control; Integrated ISR; ISR Collection Management; Knowledge Representation and Inference
