NGA is evolving its advanced analytics and modeling structure to help get the best product to the end users faster
Advanced analytics and modeling enable the discovery, assessment, understanding, measurement, enrichment, and use of geospatial data from around the world. The analytical capabilities of today’s global GEOINT enterprise continue to build a rich foundation of unstructured and structured information in multiple dimensions and formats. To address the increasing complexity of today’s challenges and prepare for future power competitions, the National Geospatial-Intelligence Agency (NGA) is investing in advanced analytical and visualization capabilities and evolving its advanced analytics and modeling structure.
Guests from NGA discussed the agency’s evolving advanced analytics and modeling structure on USGIF’s GEOConnect Series Virtual Main Stage.
“Advanced analytics and modeling gets to the core of what NGA does and who we are,” said Mark Munsell, chief technology officer, NGA. “We’re an analytic organization. We exist to take information—source material—and develop insight and intelligence from it. And then feed that to decision-makers, both strategically and tactically, to make the best decisions.”
NGA’s Modeling Landscape
NGA’s current modeling capability is primarily split between two different views—knowledge modeling and analytic modeling. In terms of knowledge modeling, NGA analysts need a discoverable and collaborative knowledge reference base of assertions against structured objects and activities to enable the synthesis of different information.
Through its advanced analytics and modeling tech focus area, NGA wants to capture an analyst’s knowledge on intelligence issues and feed it into a machine so that the machine can then help analysts to make better intelligence assessments quicker and more accurately or automatically drive their collection.
But NGA has two challenges—disparate data and knowledge sharing.
The data is either structured or unstructured and it is located in a variety of different places. Over the next 10 years, according to Rena Elliott, modeling lead, Analysis Directorate at NGA, this problem will get worse as new data continues to come in. “One of the ways that we can resolve this is to get that knowledge out of [the analyst’s] brain and into a machine so the machine can aid them in understanding the context of that data.”
With regard to knowledge sharing, when approaching a new account, analysts often get a multitude of literature to read and it takes quite some time in order to become an expert on that account. And transferring that knowledge from one person to another is often difficult. The way forward, according to Elliott, is to capture the analyst’s knowledge around the objects and the relationships they care about and to utilize existing datasets to help them build these knowledge models as an iterative process.
According to the tech focus areas, the agency’s analytic model requires analysts to conduct model-driven assessments of activities, the associated indicators, and their interrelationships across multiple domains to generate deterministic or probabilistic estimates of events.
“One of the outcomes from using analytic modeling capabilities is smart tasking, thereby enabling value-driven collection, and reducing the inundation of data that needs to be reviewed by an analyst,” said John Wall, deputy product owner for modeling, Source Directorate, NGA.
“So, it’s really a value proposition. We need to make sure that we’re collecting those datasets, smartly and efficiently.”
According to the agency’s 2020 Tech Focus Areas, NGA is investing in advanced analytical and visualization capabilities.
“We need to transition away from some traditional methods to more modern techniques,” said Kevin Ayers, deputy director, South Asia WMD Analysis Cell and senior GEOINT analyst for Analytic Modernization, NGA. “It’s really about culture change. And to make that happen, we have to actually make it easier for analysts to use the techniques, which generally require human-centered software development and not requirements-based development. It’s really a critical role that our industry partners can help with.”
As part of this tech focus area, NGA is incorporating augmented and virtual reality systems and 3D gaming technologies. These technologies have the potential to unlock new abilities for GEOINT professionals to synthesize incoming data—ever-increasing in scale and complexity—and to produce new intelligence products for NGA’s customers.
“It is a very intensive process for our analysts, working with developers, backend engineers, and data scientist, to get our capabilities to the fidelity and quality we need them to be for the level of work we provide,” said Tara Gattis, Computer Vision Lead, Analysis Directorate, NGA. “The human-centered thinking techniques we’re using now are going to be the trend over the next few years. We’ll continue to need to do that. And that’s where our partnership with both industry and academia needs to stay strong.”