Readying a new model
- By David Perera
- Oct 04, 2004
Office of Management and Budget officials have worked on the federal data reference model for more than two years, and they are preparing a final internal clearance before releasing it for a period of formal agency reviews and comments. Knowing that the data reference model (DRM) would be the federal enterprise architecture's most difficult trick, OMB officials saved it for last.
"It's going to be a lengthy process," said Bob Haycock, who left the still-vacant post of OMB chief architect in April. Agency officials plan to "lay out the model first and start to build from there" through staggered releases of three additional DRM volumes, he said.
The first volume sketches a basic data model consisting of three areas: categorization, structure and information exchange.
Categorization uses government subfunctions identified in the business reference model by classifying data in subject areas and more specific supertypes, according to a June draft DRM. The process of classifying supertypes will be addressed in future DRM documents, according to the draft.
For some time, OMB officials considered defining governmentwide processes that would establish reusable components for guiding information exchange, said Diane Reeves, a former OMB lead architect, who is now at the Interior Department.
But the first volume sets no data exchange standards, although future volumes may do so.
OMB officials apparently hope to sidestep a debilitating debate on data object definitions and instead strive for common metadata standards by creating Extensible Markup Language registries for different communities of interest.
Even basic data elements can have seemingly infinite variations. Defense Department officials once found more than 35 ways to represent the calendar date. Tagging, by contrast, seems to be a much easier way to present electronic data for sharing among agencies, Haycock said.
One purpose of the data reference model is to identify core data elements that are common across all metadata registries, a task Haycock described as Herculean. That difficulty is partly why OMB officials delayed work on the DRM until they had created the four other federal enterprise architecture models, he said.
Officials at some agencies, however, aren't waiting for OMB officials to release the model before setting up their own XML registries. "Every time you turn around, there's a new XML schema being developed for a specific application," Haycock said.
Government officials may be facing a dilemma: Plans for universal metadata standards must be made to avoid conflicting definitions. But getting agency officials to identify common data requires a degree of cross-agency data transparency.
OMB officials may tackle that paradox with emerging metadata standards in four cross-agency areas: health, defense, law enforcement and intelligence, Haycock said. "Our thinking when I was [chief architect] was, once we got the model agreed to, we would take several communities of interest that had been working
on this awhile, and we would actually begin to pile it in through those communities' development of schemas," he said.
What lies ahead is a "very painful, lengthy, difficult process," Haycock said.
Defining electronic data
A governmentwide data reference model eventually will help federal officials exchange electronic data more efficiently, Office of Management and Budget officials say. They are leading efforts to create such a model. In its first release, however, the model will not include all standards. The model may:
Categorize data according to lines of business subfunctions already identified in the federal enterprise architecture business reference model and other yet-to-be-defined supertypes.
Define three attributes for data that are consistent with international metadata standards.
The model initially will not:
Establish governmentwide data standards.
Specify mechanisms for cross-agency data exchange.
Create common Extensible Markup Language definitions.