abducting and killing Jessica Lunsford. The law that the State of Florida put into motion forces these organizations to share information. But they still don’t want to give it up, and they don’t even want someone else storing it in a data warehouse. So we pull it out and store it on a case-by-case basis.”
The key, according to Metatomix’s CTO, Howard Greenblatt, is that the ontology ties together the data according to its meaning. For each specific subsystem that the system has to deal with—say, the legacy Oracle or IBM DB2 database in a county agency’s back office—Metatomix builds an ontology representing the way that subsystem stores information. Through an advanced form of XML known as the Resource Description Framework, or RDF, those lower-level ontologies can feed information upward into higher-level ontologies that “understand” the general concepts embodied in different ways in the subsystems. “In general, we have a concept of a person and an address and a perpetrator and a booking event,” says Greenblatt.
As soon as a detainee is booked in Florida, a server programmed by Metatomix starts querying local, state, and federal databases for information corresponding to those categories and others. The data is pushed to a secure Web page accessible only by court staff and law enforcement officers; there, a single “dashboard” aggregates information such as outstanding warrants, sexual offenses, restraining orders, photographs, and biometric information like fingerprints. From the dashboard, users can drill down into individual databases for supporting data. To comply with privacy laws and with local agencies’ demands for control over the information, the pages are deleted within 24 hours after the official who requested it finishes their review.
The whole system, called the Judicial Information Exchange Model, was installed in just nine months for $2.8 million, according to Metatomix. It makes Florida the first state with a fully integrated criminal justice information system. But plenty of other states are watching Florida’s effort, according to Pilkington; similar projects are now underway in Georgia, Illinois, Oklahoma, Ohio, Texas, and Vermont. And this week Metatomix—which was founded in 2000 and has raised two rounds of venture capital, including a $7 million round last May led by Apex Venture Partners, Battelle Ventures, Dunrath Capital, and North Hill Ventures—announced that it’s an official IBM business partner in the public safety area. That means Big Blue will market the company’s semantic platform to existing customers and include Metatomix’s technology in the big, multi-vendor proposals it submits when state and federal public-safety agencies put projects out for bid.
But law enforcement is far from the only field where restricted ontologies and RDF can help officials “connect the dots,” to use the post-9/11 catch phrase. “It turns out that you can solve a number of problems with the approach that we’ve taken of leveraging both semantics and rules about how to use the data,” says Greenblatt. The company is building custom semantic-filtering systems both for financial services companies, which can use them to screen borrowers and guard against fraud and money-laundering, and for manufacturers, who can use them to make sense of disparate parts databases. Airbus, for example, turned to Metatomix for help constructing a tracking system for the millions of components that go into its passenger airplanes.
Rather than starting from scratch on that project, says Greenblatt, Metatomix is adapting an ontology already developed by NASA. And the fact that ontologies can be reused, adapted, and merged into even more general ontologies is one of the saving graces that may eventually make it possible to achieve a grand semantic Web on the scale Berners-Lee has proposed. “The issue with the semantic Web right now is that people have to agree on what the definitions are,” says Greenblatt. “There are some ontologies, like the Dublin Core for publishing and libraries, that people have already agreed to accept as fundamental. Many more will have to be worked out before we can ultimately implement ‘Web 3.0’ or the ‘Web of meaning,’ as people call it.”
But Greenblatt emphasizes that “we are not trying to build the semantic Web here—we are just trying to solve a few specific industrial, enterprise, and government problems.” And in the end, it’s this specificity that seems to be steering Metatomix toward success.
