Data Driven Activity Based Intelligence with Probalistic Knowledge Graph

In this proposal, we propose Data Driven Activity Based Intelligence with Probabilistic Knowledge Graph. address the ABI bottlenecks of current approaches by developing comprehensive methodologies to leverage available multi-INT data to extract entities and correlations to enable pattern discovery and abnormal detection of activities. Because large amount of data via persistent sensors are available, advanced processing architectures with unified representation to accommodate big data are required. We propose to leverage existing state-of-the-art technologies to incorporate Multi-INTs using Knowledge Graph representation with the lambda framework to analyze data in both batching processing and streaming processing. We seek to develop a framework and technical underpinnings for Data Driven Activity Based Intelligence and apply the framework to address challenging problems in the current multipolar military threat environment. Specifically, the proposed technologies are grounded in the advanced georeference discovery algorithm, probabilistic Knowledge graph representation and inference, hypothesis management, and pattern discovery. The goal is to provide a solution for the case of unknown-unknowns for intelligence analysis where no models or signatures are given to discover activities. The resulting technology will enable advanced intelligence analysis by allowing a more complete understanding of the Activity Based Intelligence operational picture.