Chiliad's Phase I effort will design a continuous monitoring analog to the successful retrospective concept recognition and metadata generation capability that has been deployed successfully across the FBI. Current approaches provide ad-hoc analysis with a human in the loop at each stage, with enterprise-wide data fusion; contextual analysis; concept recognition; automatic metadata generation and extraction; dynamic knowledge navigation and discovery; and task-driven collaboration across pre-indexed decentralized intelligence collections. Data are stationary, while analysis is dynamic. Unlike traditional approaches to predictive analysis driven only by historic data models, the proposed approach will combine well-understood behavior and patterns to anticipate likely courses of action, with new behavior patterns detected from live field sensor data, leveraging value from real-time analysis of `unstructured' content, deriving structured data from these sources, and leveraging further value from these structured data with triggers on relational databases. Analysis and metadata generation are executed on-the-fly, moment-to-moment using live network data, not yet indexed. Detected strategy changes from field data are fed back into the continuous monitoring loop in near real-time, improving adaptive and responsive capability that is proactive, anticipatory, and reactive at the same time. Data are moving, and profiles are stationary, but adaptive. Humans receive instant alerts based on inferential analysis, pattern prediction algorithms, and dynamic Bayesian networks, using real-time field sensor data. Response-time can be almost immediate in time-critical situations
