Feds hunt for XML core

A federal initiative launched earlier this year, called the National Information Exchange Model (NIEM), seeks to identify a core set of schemas and ensure their standardization, said Michael Daconta, the Homeland Security Department's metadata program manager.

An overabundance of Extensible Markup Language schemas could replace incompatible systems as the major barrier to information exchange, Daconta said, speaking April 28 at an ArchitecturePlus Seminar in Washington, D.C.

As agencies begin implementing service-oriented architectures, which depend on XML for data exchange, NIEM officials want to expand the accepted set of common schemas by finding more participants, Daconta said. Environmental Protection Agency officials have been briefed, he added. So far, the Justice, Homeland Security, Transportation departments and the National Institute of Standards and Technology have signed up for NIEM. A project Web site became active earlier this week, he said.

The project seeks to identify two categories of common XML schemas: a universal core shared by every information domain and more specialized overlapping schemas specific to just some of the domains.

The basic assumption, is that "any time two information domains intersect, that is a piece of core data," Daconta said. "You don’t totally control it."

Taking a realistic approach to identifying areas of data overlap will make core schemas possible, he said. Organizations contain more data than needs standardization, Daconta said. "The scope of this is information you either share today or are confident -- say more than 90 percent confident -- that you're going to share tomorrow," he said.

The universal core of schemas that cut across all domains will be relatively small, from 100 to 200 types, Daconta said.

The process of joining NIEM will require staggered harmonization of their schemas with core schemas, Daconta said.

Once the number of universal core schemas stabilizes, NIEM officials will consider submitting them to an international standards body for codification, he added.

Harmonizing the schemas will get rid of pointless metadata discrepancies. "Let's be honest, human nature has the 'not-invented-here syndrome,' and we like to debate and argue over trivial things," Daconta said. "You've got to get the point where you discover what are the artificial differences vs. what are the real differences."