ARAKNI: Adversary Reasoning Assembling Knowledge Networks with Integrity

SIFT proposes to build ARAKNI (Adversary Reasoning Assembling Knowledge Networks with Integrity) to innovate red-force behavior inference with integrity, aiming to infer hierarchical red-team intent and plans from raw, uncorrelated data and providing meaning-making interfaces for human interpretation and confidence assessment.  A central research challenge is to jointly (1) infer red-force plans and intent from observable data with plan recognition algorithms, and then (2) utilize these inferred plans and intent to predict future (or past) red-force actions, which may serve as collection goals to further reduce uncertainty.  To address this challenge, ARAKNI will utilize SIFT's ELEXIR highly-scalable plan recognition tool that infers and tracks likelihood-ranked explanations of red-force behavior across world states, allowing ARAKNI to prioritize and compare alternative hypotheses of red-force behavior.  Further, ARAKNI will enhance ELEXIR with projective plan recognition to interleave plan generation within its plan recognition, allowing ARAKNI to make grounded, verifiable predictions about previous and future states in a partially-observable world.  Finally, ARAKNI will utilize SIFT's AFRL SBIR-funded human-machine workspace to allow analysts to interactively explore machine inferences for sensitivity, likelihood, diversity of sourcing, assumptions, information gaps, and estimated confidence, supporting analytic tradecraft and meaning-making.  SIFT is well-positioned with transition customers in the DoD to build ARAKNI and deliver it to analysts who need it.