Calais are concentrating upon what we refer to as "izza's": New York izza State, Microsoft izza company, etc. This is an essential element of effective Semantic Web tagging, and is known, in the business, as "Named Entity Recognition." After all, if you can't tell that Proctor and Gamble is a company, not just two guys, you don't have a very valid place to start understanding the content that refers to P&G (it's nice, also, to know that P&G is Proctor and Gamble: easy for a human, tough for a computer.)
The release, in the latest version of the OpenCalais service, of so-called "Social Tags", is a solid step forward. They're employing a taxonomical approach that permits the creation of "meaningful" classifications of named entities, such as "sports cars", which publishers will find both relevant and monetizable. Bravo!
To me, this represents a solid step forward because it recognizes that classification schemes must support the pragmatic goals of those who apply them. This resonates strongly with the OpenAmplify view that forward motion in the Semantic Web must come through real-world applications.
I like what these folks are doing.
Mike Petit | Co-founder/CIO | OpenAmplify