Bootstrapping the semantic Web

It's tempting to draw parallels between the careers of Albert Einstein and Tim Berners-Lee. Both men made world-transforming breakthroughs and then pursued even grander visions. Einstein, of course, never found the unified theory he sought for three decades. A lot of people think Berners-Lee's vision of a semantic Web will prove equally elusive.

We can all imagine the desired outcome: a version of the Web where items are related explicitly, not merely by co-occurrence of words. But skepticism has greeted the "semweb" technologies that Berners-Lee has been spearheading in the W3C. The approach is based on what's called "ontology," which the W3C defines as "a representation of terms and their interrelationships." Critics argue that we'll never agree on (or consistently apply) an ontology -- and they point to Google as proof that we don't need to.

Two companies I've encountered recently think there's a middle ground. One is Semagix, which offers an application toolkit, called Freedom, to help developers create, and then build on, a domain-specific ontology. Take the case of an anti-money-laundering application. The ontology is derived from authoritative information about individuals and companies provided by the likes of Dun & Bradstreet and Hoover's. Given such a framework, automatic classifiers can read unstructured documents -- e-mail, news feeds, Web pages -- and attach them to the framework. As a result, Semagix says, you can answer questions such as, "Which recent news reports mention companies that share directors with company X"?

Digital Harbor is taking a similar approach. The company's PiiE (Professional Interactive Information Environment), originally pitched as a rich Internet app-dev toolkit, has lately shifted toward "business ontology." Digital Harbor's Fusion Server helps developers define a set of terms and relationships, populate that framework with structured data, and then attach unstructured data to it. And as does Semagix, Digital Harbor emphasizes fusing ontology and data so that users can "connect the dots."

Of course, it's never as easy as we'd like to imagine. Consider Eliyon, a company that's gathered public information about more than 22 million people to support sales, recruiting, and other applications. As it turns out, I am several of those people. In addition to my current title, InfoWorld Test Center lead analyst, I show up as executive editor of Byte Magazine and contributor to Linux Magazine. And while those were once accurate descriptions of me, I have never been a member of Blue Titan's board of advisors, and I am not the inventor of RSS.

It's true I could register with the site, coalesce my correct identities, and purge the wrong ones. But authenticating with a credit card in order to update a profile that Eliyon owns is a nonstarter for me. Back in June, on my Weblog, I suggested the alternative that would suit me: I'll maintain my own profile on the Web and syndicate my data to anyone who needs it.

Semantic-Web naysayers think people and organizations can't be bothered to assert machine-readable facts about themselves. And, today, that is undoubtedly true. But when others assert facts about you -- as they increasingly will -- the tide could begin to turn. Individual acts of self-defense may ultimately combine to bootstrap the semantic Web.