DDABI: Data Driven Activity Based Intelligence with Dynamic Knowledge Graph

In the current asymmetric threat environment, it is critical for the DoD intelligence community to develop ability to automatically integrate information and data from Multi-INT sources in a complex environment to produce a dynamic, comprehensive, and accurate activity based intelligence. In particular, there is a need to develop a proper big data enabled analytic system by integrating and associating multisource data such as unstructured text, geospatial data, and open-source intelligence with scalable architecture, in order to shorten the critical data collection-to-analysis timeline. Specifically, new approaches are needed to bridge the gap from individual exploitation to analysis-based tradecraft to provide timely and actionable decision-relevant information.This effort proposes to develop comprehensive methodologies to leverage available multi-INT data to extract entities and their correlations to enable pattern discovery and anomaly detection. To deal with the large amount of data via persistent sensors, an advanced processing architecture is proposed with unified representation to accommodate big streaming data. Specifically, the proposed technologies are grounded in dynamic knowledge graph representation and inference, spatial-temporal reasoning for pattern discovery and anomaly detection, and advanced scalable computational architecture. The goal is to provide a solution for intelligence analysis where no models or signatures are given to discover activities.