Governmentwide adoption of an Extensible Markup Language schema and ontological reasoning could turn the federal enterprise architecture into an operational architecture.
Turning the federal enterprise architecture into an operational architecture would be no small achievement. Completion of the data reference model is one of the first necessary steps, and a CIO Council task force is working to make it a tool for describing the structure, categorization and exchange of agency information.
By constructing a governmentwide schema based on Extensible Markup Language, the task force is knocking down barriers to data sharing. Although the approach has critics, the government's first priority is to move forward with a good solution rather than wait for a perfect one, said Michael Daconta, the task force’s leader.
Rather than implement a single data reference model, as originally proposed, the group has decided to implement it in parts based on different levels of specificity. The task force is debating the merits of a top-down approach vs. a bottom-up one for establishing what data is included in the most general implementation.
One goal is for federal organizations with common interests to be able to add new attributes and entities that are relevant to their community to the data reference model schema. For example, the health care community could require the data element of "person" to carry attributes for blood type. "No matter which approach is taken, we are certainly going from the general to the specific, so the highest levels [of the data reference model] are the more general and then the details are left up to communities of interest," Daconta said.
The task force has postponed the second release of the data reference model until it can test the concept among voluntary agencies. An intermediate version will be ready Oct. 17, Daconta said. "During the test period, it's still going to be a draft.… It will not be mandated because we have agreed to test it first." The Homeland Security Department and Environmental Protection Agency are likely candidates for the test programs.
However, an operational data reference model is a necessary but not fully sufficient condition for an operational federal enterprise architecture, Daconta said.
"Data is just one part of an overall enterprise architecture,” he said. “You have to include the services piece. You would have to make the service reference model actionable, too."
With an operational federal enterprise architecture, agencies would share and reuse common services. "Is it the goal of the service reference model to enable service invocation between agencies? If the answer is, ‘Yes,’ then you're at a whole another level of specificity," Daconta said.
A fair amount of reference model reconstruction may be necessary before they can be connected, said Ira Grossman, chairman of the Chief Architect’s Forum. "That's part of the maturity process."
But proponents of a cutting-edge solution say they know how the reference models could become an operational architecture.
Real-world relationships are more complicated than expressions in the hierarchical rankings of taxonomies, said Brand Niemann, co-chairman of the CIO Council’s Semantic Interoperability Community of Practice. Taxonomies by themselves are limited, but they can be greatly enhanced through an ontology model, Niemann said.
Ontologies can find relationships among agency architectures constructed using different vocabularies and tools, said Roy Roebuck, chief architect of the Continuity Communications Enterprise Architecture Program Office. I say tomato, you say tomate, but although I speak English and you speak Spanish, we're both talking about a round red fruit. An ontology, "serves as a translator between these different organization, functional, technical domains," Roebuck said. After establishing common concepts, the platform is set for a much richer set of relationship categorization among entities.
Unlike taxonomies, which categorize relationships in hierarchies, such as "the tomato species is a child of the lycopersicon genus," ontologies allow for more complex relationships. An ontology can categorize elements with relationships such as full or partial equivalence, reference, contingency, sequence and containment. Computers can quantify and understand direct relationships by using ontologies, which can lead to the discovery of previously obscured relationships among datasets.
Making ontologies computer-processable also ensures consistency. Two architects could develop two separate line-of-sight relationships based on data from the reference models. "That's where uncertainty comes in," Niemann said.
But an ontological model extracts concepts, defines relationships and allows machine crunching to consistently spot hidden connections. "Now you have something you can actually use to build real applications," Niemann added. Ontology has the potential to make the federal enterprise architecture make a quantum leap forward, Roebuck said. "It's a revolutionary approach. But it's revolutionary in that it's doing nothing more than tying together what's already there." Architecture application users would never need to know about the ontological wiring hiding underneath their dashboard.
To users, a federal enterprise architecture application "is like looking under the hood of a Lexus," Grossman said. "There's a lot of complexity under there and a lot of microprocessors, far more than we can understand. But it's there, and the car runs, and we know how to drive the car."
NEXT STORY: Unhappy at FEMA?