The Air Force Research Laboratory (AFRL) awarded Charles River Analytics, a developer of intelligent systems solutions, a contract to develop a system for the Assessment of Adversary Capability and Capacity via Intelligence Analysis (ACACIA). “The goal is to help analysts in the Air Operations Center (AOC) or Joint Information Operations Center (JIOC) gain an understanding of what an adversary is capable of doing,” said Dr. Avi Pfeffer, Principal Scientist at Charles River.
Air Force intelligence analysts attempt to anticipate an adversary’s intent, capabilities, and likely actions. They analyze the analyst’s available offensive and defensive systems, state of readiness, training, and any contextual features (such as weather conditions) that might affect the adversary’s capabilities. ACACIA aids in this process by helping analysts track and integrate these factors. ACACIA also incorporates expert knowledge that it can use in analyzing the current adversary state.
This expert knowledge is especially useful for junior analysts, who may be new to the field or inexperienced in intelligence analysis. For instance, when a junior analyst questions the ACACIA system, the system will analyze the available data, present the relevant analysis factors to the junior analyst, and provide an initial assessment of the adversary’s capabilities and intentions. The system provides this analysis based on a probabilistic model which can provide expert-level reasoning about missing and incorrect information (which, in turn, may lead to requests for additional or confirming data). The analyst can also modify the input data stream to assess the sensitivity of the analysis results to the underlying data. For instance, will the assessment change if predicted weather conditions differ from those assumed for the analysis?
Making this all possible is a probabilistic model-based ”engine” that uses the probabilistic programming language Figaro. Figaro makes it possible to rapidly construct probabilistic models for a wide variety of situations from a common knowledge base. ACACIA is one of Charles River’s projects using probabilistic reasoning and modeling. Another is the Probabilistic and Relational Inferences in Dynamic Environments (PRIDE) system.