Jennifer Zaino
SemanticWeb.com Contributor

Recovery.gov, the U.S. government web site with data on spending related to the Recovery Act, re-launched this week. The re-design, a project of geographic information systems vendor ESRI, lets users navigate through maps to trace how funds are getting spent in each state, and download the data on federal contracts, grants and loans that was rolled up from across the states in both Excel and XML formats.

One thing that Diane Mueller, vice chair of the XBRL International Steering Committee and VP, XBRL Development, at JustSystems, likes is that the effort supports a number of industry standards, which helps users avoid repeating mapping exercises every time they download data. The next step is creating ontologies or maps that make it easier to make connections across different types of data sets -- making it a more seamless process, for example, to tie geospatial information related to an agency such as the Department of the Interior to other agency data about funding that is being applied to saving animals in western Utah.

"That's when we get into conversations like XBRL for NIEM (National Information Exchange Model), one of the fine federal standards for sharing information between federal agencies," Mueller says. "You have to create those mappings between the standards as well. That's the next level to go on."

Such are the challenges around really opening up government data for use by the citizenry. The ability to make government data easily accessible is just the first step in creating real accountability -- it's also key to make sure that users are able to access and interpret that data accurately. Some of the U.S. government projects fall short there, with Mueller pointing to the recent news that the White House is disclosing visitor access records in the .CSV spreadsheet format.

"This is a good example of how not to post your data," she says. After you pull that .CSV file into Excel to see who visited whom on what day, maybe you want to make some synaptic semantic connections between that visitor and the company he works for, for instance, so you trot over to the SEC site and grab the corporate filing information in XBRL. And then maybe you want to find recent news around that executive and his company, and that requires leveraging news feeds conforming to the NewsML format. Then you start cutting and pasting all these data finds together-- a manual and so in and of itself an error-prone process -- for a mash-up and you lose the metadata associated with it. Now its validity can't be verified by others with whom you'd like to share the information, because they can't directly link back to the source for an authenticity check.

"The thing that always is missing in most of these conversations about publicly accessible information is harmonization," she says. "The very first thing you have to have is the metadata."
Applying industry standards like XBRL for financial data or NIEM opens the door to knowing from one application to another how to accurately interpret information. It would be useful if government agencies at the federal, state and municipal level were to make their financial reports available in XBRL, as corporations must do for the SEC, she thinks, to get some real benefits around tracking how various agencies are spending their money, what their balance sheets are, and so on. But that will take a lot of work, and a whole other taxonomy effort as the current XBRL taxonomy set is for US GAAP for corporate accounting, not the GASB government agency standard.

Mueller expects issues around such concerns will be addressed during next week's Workshop on Improving Access to Financial Data on the Web in Arlington, VA, which is co-organized by the W3C and XBRM International Inc. and hosted by the FDIC. She is chairing the W3C XBRL workshop there, and expects one of the topics of conversation to also cover problems converting XBRL data to RDF, so that you get use semantic tools to query against the data.

"I guess a lot of what we'll see in the future on sites like Recovery.gov will be RDF," she says.