Open Source Information Tactical Exploitation (ONSITE)

We propose to design and demonstrate the ONSITE Natural Language Processing technology. ONSITE uses innovative approach to NLP focused on achieving state of the art processing throughput in the presence of noisy input data which achieving semantically rich extractions from unstructured open source text to support tactical applications requiring accelerated operational timelines. ONSITE seeks to strike the correct balance in the design of NLP algorithms for fast execution on a single laptop computer while still providing useful semantically rich extractions. Floating-point and math related operations are limited in favor of optimized heuristics leading to higher rates of CPU throughput and at the same time supporting the structuring of calculations to allow increased CPU cache utilization. ONSITE incorporates the use of specially constructed ontologies to increase processing throughput and increased semantic resolution of extractions. The initial ONSITE application proposed by KBSI is influence operations targeted at enemy networks both physical (existing in the real world) and virtual (existing online). However, the set of supported tactical applications provided by ONSITE may be extended to different tactical application areas including patrol mission planning, insurgent extraction, hostage rescue, and targeting.