The Office of Inspector General (OIG) of the Department of Health and Human Services (HHS) is building on its AI text analytics capabilities and creating a new “bring your own documents” web portal that will allow clients without analytics experience apply natural language processing methods to an ad hoc set of documents, according to an HHS official.
Keith Bocian, senior data scientist at HHS OIG, explained at an FCW event on April 13 that text analytics is an emerging new technology the agency is looking to develop. Bocian said his office came up with the idea for a portal after partnering with business owners around HHS to understand “their potential use cases and get their advice” on where they should. prioritize their AI efforts.
“They really asked us to focus on opportunities in the area of text analytics, allowing customers with no background in analytics to apply natural language processing methods to ad hoc sets of documents and then also identify multiple opportunities in the field of automation,” said Bocian.
“Key to this process was our partnership with business owners,” he added. “They really drove the prioritization and allocation of resources against these different use cases. I think it’s so important to have the mission, the business needs, to drive the allocation of resources towards technology projects, rather than doing technology for technology’s sake.
The ultimate goal, Bocian said, is to create a “bring your own documents web portal” that can enable users without AI or analytics experience to search for documents, extract relevant information, and be able to to use tools such as entity recognition and topic modeling to understand who and what the documents are about.
Bocian said HHS OIG is still developing the portal, but plans to release “limited elements” of the web portal for users in a test environment later this month.
He said the agency has use cases for the portal in the areas of hotline and whistleblower complaints, in addition to ad hoc “bring your own documents” use cases.
“People can bring an email inbox and be able to understand over time how the content of those messages has changed,” Bocian said. “So, [we’re] really looking to enable text analytics for a broader set of use cases than was previously possible through the use of natural language processing, and make them accessible in a web portal for users no analytical experience.