Feds take (baby) steps to fight fraud with analytics

Government agencies have begun, tentatively, to mine their vast stores of data for the public good.

By Tam Harbert

Computerworld |

Previous 1 2 3 Page 3

Page 3 of 3

Stumbling blocks

There are some efforts to spread the knowledge and capabilities learned from individual projects more broadly across the government. The RATB, for example, has started providing fraud-prevention services to other agencies through several pilot programs.

However, by law the RATB is supposed to "sunset" on September 30, 2013. The Obama administration has created another board -- called the Government Accountability and Transparency Board (GATB) -- to study RATB's system and processes and consider how they might be continued and used governmentwide.

However, the GATB "is just paper," says Hudson Hollister, executive director of the Data Transparency Coalition, a trade association of technology companies and nonprofit organizations that advocates for data reform in the U.S. government.

"It's just a group of people studying the issue and making recommendations." If no specific entity is designated to take over RATB's work soon, then RATB's system, contractors and expertise will be lost, he warns. Although proposed legislation, called the DATA Act, could ensure that wouldn't happen, it has yet to pass Congress.

The AGA report notes that the systems developed by the United States Postal Service (see USPS fights fraud four ways) could be easily applied by other agencies. "Every agency uses contractors and has workers' comp claims," says Steve Sossei, co-researcher and co-author of the report, "so why don't we make sure that all agencies collect this information and analyze it in the same way?"

There's willingness in the agencies to share their capabilities with other federal organizations, says the AGA report, "however, there is no clear plan to leverage the government's investment in analytics."

Wanted (desperately): Standards for government data

In May 2012, the Obama administration rolled out its digital government strategy with a goal of "unlock[ing] the power of government data to spur innovation across our nation and improve the quality of services for the American people."

When it comes to data analytics projects, however, federal agencies are so far falling far short of that lofty goal, observers complain.

Much of what passes for analytics in government is nothing more than automating what was previously done, or should have been done, by human beings. Things like verifying information by cross-checking sources, for example. "It's not analytics. It's search," says Hudson Hollister, executive director of the Data Transparency Coalition, a trade association of technology companies and nonprofit organizations that advocates for data reform in the U.S. government.

There are several reasons why government is not further along in applying analytics, says Hollister. One is the sheer bureaucracy, which makes communication and coordination among agencies difficult. Perhaps more important are various legal barriers to inter-agency sharing of information.

The Computer Matching and Privacy Protection Act, to cite just one of many examples, requires agencies to have separate written agreements in place with each entity with which they wish to share data. While the government needs to protect the data of private citizens, there is plenty of data that could and should be more easily shared, says Hollister.

For example, company financial data filed with the U.S. Securities & Exchange Commission could be matched with applications for government grants or loans. If that had been done in the case of troubled solar-panel manufacturer Solyndra, the government might have avoided a bad loan and subsequent scandal, he says.

"The privacy issue pertains mostly to individuals," he notes. "Agencies, companies and entities that get federal grants and contracts have no right to privacy."

But even if laws are changed and the government somehow sheds its bureaucratic lethargy, a lack of standards would still inhibit even simple data analytics projects, he notes.

Today, agencies keep records about financial transactions like grants and contracts using a mishmash of different systems and data formats. In fact, the agencies don't even use the same formats to identify each other. "For example, the Department of the Treasury and the Office of Management and Budget use competing and incompatible codes to identify the federal agencies and their subdivisions and bureaus," says Hollister.

While there is growing recognition that government should do more with its data, Hollister says, right now, "the government just isn't able to do as much with data as the private sector is."

Frequent Computerworld contributor Tam Harbert is a Washington, D.C.-based writer specializing in technology, business and public policy.

Previous 1 2 3 Page 3

Page 3 of 3

Bing’s AI chatbot came to work for me. I had to fire it.