FBI’s data chief seeks better analytics as agency moves to cloud next year

National security agencies collect hordes of data every day. But drilling down into that information and getting valuable analysis poses a whole different set of challenges.

While several agencies’ offices have developed promising test cases for artificial intelligence, machine learning and automation tools, Robert Osgood, a former computer forensics examiner at the FBI, said the government has hardly scratched the surface when it comes to leveraging these tools.

“The big problem with these type of methodology platforms is that we don’t really know how to effectively drive data through them,” Osgood, now a professor of data analytics and computer forensics, said Thursday at AFCEA Bethesda’s Data Analytics Breakfast.

Maria Voreh, the FBI’s chief data officer, said the bureau has so much data that it is “near impossible” for the agency to get value out of it quickly enough.

Advertisement

“The amount of data I’m getting in, and the amount of sense that we can make of it, there’s a huge gap delta, and that’s our data debt,” Voreh said. “My job is to make sense and reduce that data gap. Whatever I have to do to organize the data, share the data within our agency or other agencies, I’ve got to reduce that gap, or else I can’t serve the mission.”

The FBI plans to move to the cloud within the next year in order to share its data both within the department and across other agencies.

“To be able to share broadly with other partners or to meet a joint U.S. mission, we’ve got to move our data to the cloud,” Voreh said. “These siloed, on-prem systems aren’t going to cut it. Our budgets aren’t getting any better, so buying more computers and putting them into a warehouse somewhere isn’t going to happen. We’ve got to invest in the cloud, we’ve got to move our critical systems so that we can get the data out of them.”

While the move to the cloud will take the FBI multiple years, Voreh said the migration will allow the agency to access all of its data more quickly.

“It’s going to take time,” she said. “In five years, we’ll probably get a lot further along because once we break the seal of the security, the protections,  compliance and the comfort of the social understanding within the agency, then it becomes easier. It’s always that first piece going to a new technology that’s the hardest.”

The FBI put out a request for information (RFI) for commercial cloud services on Feb. 16.

Obtaining reliable data analysis is critical for DHS components like Customs and Border Protection, which in fiscal 2017 screened more than 400 million people entering the United States — or more than 750 people per minute.

David Bottom, the chief information officer of the Homeland Security Department’s Office of Intelligence and Analysis, said getting the agency culture to embrace new tools plays a major role in getting better data analysis.

For those CBP agents on the border screening 750 people a minute, Bottom said it is easy for them to understand the value proposition of better analytics.

“As we approach the cultural issue, you’ve got to get to people where they are,” Bottom said. “Answering the question of what’s the return on investment is a good first step. People recognize that there is a ‘there’ there — that there is a value. Answering what that is, I think, where we are right now.”

When it comes to data analytics, Voreh said the FBI has “pockets of excellence,” across its 56 field offices that accommodate more than 35,000 total employees. But getting everyone on the same page with data literacy, she said, has been a challenge.

“Not only do I have the technical challenge of how do I get that data to a central location that people can share, which is a  challenge,” Voreh said. “It is also how can I change the mindset and educate those that make our policies,” including legal advisers and agents in the field.

Selling data analytics as a long-term money-saver helps, Bottom said, especially at a time when constrained budget force agencies to make tough choices.

“Honestly, it comes down to a case of, ‘Well, do I want a national security cutter,’  if you’re the Coast Guard, ‘Or do I really want to invest in data standards?'” Bottom said. “You have to have some value proposition there.”