A new challenge to automate one of the most tedious jobs in government


Best listening encounter is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s day-to-day audio interviews on Apple Podcasts or PodcastOne.

A lot of the information that the governing administration creates demands to be rated harmless to distribute, managed but unclassified, or perhaps secret and categorized. That’s simplifying this big but never-ending task. Now the Defense Department has introduced a problem prize software to create an artificial intelligence tactic to automating some of this tedious task. Federal Generate with Tom Temin  bought the facts from Doris Tung, the acquisition division manager in the Philadelphia division of the Naval Surface Warfare Heart.

Tom Temin: Miss out on Tung, very good to have you on.

Doris Tung: Certainly, thank you for having me.

Tom Temin: And you are seeking for a procedure to establish, I guess, the CUI, the managed but unclassified info. Let us begin with that style of facts. Is that the toughest to discover or the most subtle, or why commence there?

Doris Tung: Properly, the managed unclassified data, which I’m likely to refer to as CUI, it’s hard to mark due to the fact it has in excess of 120 types, and there are subsets of those people. So for an stop consumer to detect no matter whether your document demands a distinctive marking or not, can be quite cumbersome. Whilst with labeled paperwork, you are really positive no matter whether or not you’re doing the job on a plan that’s heading to be, you know, solution, top rated key, and the files that are generated from that want to have their suitable marking. So CUI has been all over as a requirement for awhile. But since of its huge the vast majority of categories, and specific marking necessities, and also legacy markings with “for official use only”, and matters like that, it can get complex for a consumer to figure out whether a doc is CUI, and then “how do I mark it?”.

Tom Temin: And I picture, there’s a excellent possibility for inconsistency from man or woman to individual or unit to device or bureau to bureau, much too?

Doris Tung: Oh, unquestionably. Believe about all the documents that we generate in the federal government. We’re developing so numerous documents, particularly electronically now, much too. So you know, absolutely everyone is building their personal selections on whether it requirements to be marked, and then doing it properly. Simply because there’s pretty precise necessities on what do. You need to set on the header or the footer of the document. And then if you’re accomplishing email messages, you know, how do you distribute CUI? So there is distinct needs that an end-user from individual to person could not be mindful, and they’re just making use of what they think is appropriate.

Tom Temin: And before we get into the information of the problem you’ve introduced, why is it coming via the Philadelphia division of the Naval Surface Warfare Middle of all the possible locations in the Navy?

Doris Tung: I’m a section of a division of Navy management program termed “Bridging the Gap,” a development method for concentrating on expanding senior govt assistance. And so as component of this software, senior government service from the Navy participates by offering true lifetime issues for the workforce to resolve so we can do some action understanding. And Mr. Alonzie Scott, who is a SCS at the Workplace of Naval Analysis, he offered his dilemma to this application and our workforce,. You know, I’m coming out of Philadelphia, he introduced a obstacle of, you know, how do we simplify marking of controlled unclassified information and facts utilizing and leveraging automation and artificial intelligence and device studying? I work in the contracts section, and I’m a contracting officer, and as section of the Naval Area Warfare Heart, we do have the authority to challenge prize worries. And that was a remedy that our team came upon. You know, the crew customers consist of people across the Navy.

Tom Temin: Safe and sound to say the output of this task could have Navy-large implications, although, or even DoD huge.

Doris Tung: Appropriate, correct. Unquestionably. I indicate, I consider it could go over and above DoD, simply because we did have discussions as portion of our marketplace analysis with compact business administration, protection technical details heart. And you know, persons are all struggling to determine out how do you properly employ this where the buyers comprehend how to mark it, and it’s possible getting off some of that load off of the close-consumer. So it could have probable implications for Navy and potentially beyond.

Tom Temin: We’re speaking with Doris Tang. She’s acquisition division manager in the Philadelphia division of the Naval Surface Warfare Heart. Tell us about the problem, then, this is a not a grant software, but a prize challenge-style of method. And who are you achieving out to? And what are you hoping to appear up with?

Doris Tung: So the prize obstacle we determined to go with this process vs . any standard Considerably, you know, Federal Acquisition Regulation-primarily based contracting, since the prize problem lets us go out to the community. So it can be organizations, nonprofits, individuals, anyone can take part. There is particular restrictions, but typically, you know, any individual who has a alternative can submit their concept. So the prize challenge is to question if any one has a solution in which they can leverage the artificial intelligence device understanding to automate the marking of the document, and we have broken up the obstacle into two phases. In period one, which really just closed, is a white paper to exhibit, you know, what is their prototype, and then they will have a down pick out, where by we move on to phase two. And all those individuals then can then actually create a prototype, and then we’ll take a look at it with genuine documentation to see if they can mark it properly. And the winner that will be selected, you know, would have the highest precision price, so we’re enthusiastic to see what remedies does market and the general public have to solving this dilemma?

Tom Temin: And do you have some objective sets of paperwork that everybody has agreed these are certainly CUI, mainly because earlier, we talked about the variability that can come in there. And you outlined 120 possible classes. And we’ve read this for a lot of yrs about how lots of layers there are. So what’s your reference kind of facts?

Doris Tung: So for the prize challenge, we are focusing on offering just a subset of the CUI classification. So focusing on the privacy and the procurement and law enforcement. So we have documentation that we know for sure is marked accurately. And there’s a sample set, you know, with artificial intelligence and equipment learning, the a lot more files that you can see the resource, the a lot more the machine can master. So they will need the information. So we recognize that section of this is that we will need to give them a superior details set for the software to seriously master. So we’ve sort of been scrubbing as component of our workforce, producing these documents, making certain that it is harmless to share with the public as very well, for this challenge. But we are concentrating on just selected subsets and then hopefully, you know, based on what is the consequence of this prize challenge, then, you know, increasing beyond just these sure subsets.

Tom Temin: And do you also have patently un-CUI that you throw in there to kind of strain exam the algorithm, for case in point, like throwing in a comic ebook or a novel?

Doris Tung: I suggest, we surely considered that. We do have non-CUI paperwork so that the resource can master what is CUI and what is not CUI. But that is a very good notion about throwing in a comic reserve. That is something we’ll have to take into consideration.

Tom Temin: And I was just wondering if the algorithm can also location categorized by accident that could get in there. That would be a function, I assume you would want to have like a purple light-weight comes on and claims, “Hey, wait around a moment, this is not only not unclassified, but it ought to be labeled.”

Doris Tung: Oh, that would be an excellent enhancement for the instrument. Appropriate now we’re only concentrating on just can it even determine out is it CUI, non-CUI? And then, you know, if persons have an capacity to even tackle that aspect, we would really like to see if they included that labeled piece, because categorised is also a piece. It could be CUI and labeled. So there are actually a whole lot of variability to documents that you know, after hopefully, we can even just resolve this fundamental dilemma, then we can then transfer on to see what form of prospective these applications could have. That would be some thing I imagine folks would want.

Tom Temin: And what do you suspect are some of the techniques that this could be carried out by? For case in point, is it a straightforward word research and look at sort of thing? Or is it extra complex than that. Is there context? Is there syntax? Simply because you’re dealing with mainly published paperwork honest to say?

Doris Tung: That is good to say that it is all prepared paperwork. So we did discover what software program was present out there. And there are equipment out there now with creating CUI marking software with search phrase searches. But we discovered that to be problematic, for the reason that you’re likely to count on an individual attempting to recognize all the keywords and phrases that could probably flag a specified category. And so we’re speaking about 120 types, and then there is a subset. So do we have men and women who are capable to genuinely hone in on what search phrases would flag every of all those categories? So that’s why we shift towards the equipment finding out to artificial intelligence device understanding when the equipment then reads all these info sets, then it can figure out, you know, which of these words and phrases are, you know, I necessarily mean, that is the section wherever we’re hoping that the individuals not the prize challenge is going to inform us like how can your device do this?

Tom Temin: And now you are finding the white papers in, what is the following phase? And does this turn out to be anything that as a technologies transfer prospect or a thing, you would convert into a product or service that the Navy could buy?

Doris Tung: So the following stage right after we evaluate the white papers is the tech demo. And probably, what we’re hunting into is, you know, there is new procurement cars and approaches, these as the other transaction agreements out there. So we are on the lookout into, you know, primarily based on the accomplishment of phase two, wherever they do the demonstrations, we will then pursue no matter if it is really likely to be anything like a product or service that we can truly procure, or no matter whether there just wants to be additional abide by-up procurement solutions to see. Because there are other Navy, Marine Corps functioning technique necessities that we also have to take into account that proper now the obstacle isn’t genuinely restricting the participants in that manner yet.

Tom Temin: Positive. So when you get this solved, possibly you can consider on deal creating.

Doris Tung: Yes, I assume it would definitely be a scenario that could definitely be delved into.

Tom Temin: You know, I assure you’d rocket to the SES if it received that a person solved. Doris Tung is the acquisition division supervisor in the Philadelphia division of the Naval Floor Warfare Heart. Many thanks so significantly.



Resource by [author_name]