Intelligent Enterprise

There couldn't be a greater contrast between the views on semantics of Web creator Tim Berners-Lee and of data warehousing figure Bill Inmon.

For Berners-Lee, the value of semantics is so clear that it no longer has to be justified. As reported by eWeek's Jim Rapoza, for Berners-Lee, only standards adoption stands in the way of Semantic Web realization and the vision of automated network computing that Berners-Lee has been articulating for a decade now. Bill Inmon, by contrast, questions the fundamental concepts that underlie Berners-Lee's vision and are key to enterprise advances that range from Master Data Management to search-enabled BI. Inmon writes,

"When it comes to semantics, I don't just get it. You can call me misguided, an old fuddy duddy, or just plain dumb. In one way or another, perhaps all of those names fit. But at the end of the day, I just don't understand semantics."

What I don't get is how Inmon can miss something so obvious. Type "2+2" or "map wisconsin" or "bill inmo" into Google and you'll see semantics at work, and that's just easy stuff. (Inmon's complaint, by the way, didn't stop him from presenting on self-organizing maps, which rely completely on extracted semantics to generate graphical representations of information classifications, at the 2006 Text Analytics Summit.)

I also don't understand how Berners-Lee's thinking could have become so distant from business reality. eWeek's Rapoza reports that "Berners-Lee said that there's one simple way to determine if a product is actually a Semantic Web technology: Look for the standards support. If the product doesn't support core standards such as RDF, OWL or SPARQL, then it isn't a Semantic Web product." The reality is that only a minority of Web products will ever meet these criteria for standards support and only a tiny proportion of Web content will ever be neatly marked up according to such standards. "Folksonomies" and text analytics are examples of approaches to generating semantics that have emerged to fill the gap, but the best example is the one I provided in discussing Bill Inmon: semantically enriched search. Believe me, Google doesn't rely on information-publisher use of RDF or OWL to respond to "map wisconsin" with maps of Wisconsin.

How did we come to this pass, where a visionary confuses semantic technology with fundamental concepts and where someone with Inmon's profile states that he sees no value in any part of semantics, in the technology or the concept? Fortunately, there's much, much more to Berners-Lee and the World Wide Web Consortium (W3C) he directs than the quotation about standards suggests. And the majority of IT managers and practitioners do get the value of semantics despite Inmon's protests. Nonetheless, these two, contrasting views do provide helpful reminders. Standards are important but not always essential, and computing progress doesn't always follow the directions we think it should or even that we fully understand.