Sunday, May 20, 2007

Linked Data - a new name for RDF..

Aside from the collaborative construction of knowledge workshop that I discussed in my last post, no particular session or presentation from WWW2007 really stands out in my mind. That being said, the various sessions organized under the rather loose term "linked data" seemed to me to offer the most exciting new developments. Whether this was because Tim Berners-Lee was at most of these sessions or whether he was at these sessions because this was where the action was, I can't be sure.

The concept of "linked data" as far as I can tell, is essentially the same as web-based data integration which, IMO, is essentially the applied (non-theoretical/"actually useful") version of the semantic web. The basic ideas (described here) are basically that data should be represented in RDF and should be exposed on the Web such that the URIs can be "linked to" by other RDF statements in a similar way to how HTML pages can be linked to from other pages (but with the additional semantics provided by the RDF language). As I thought this was the basic idea of RDF all along, I'm a little uncertain why the new name came into being - perhaps its just easier to get across to people.

Anyway, the point is that, all of a sudden, there seems to be a whole lot of linked data/resolvable RDF out on the Web (finally!). One of the most exciting academic projects under this umbrella is DBpedia. DBpedia has converted the structured information present in Wikipedia into a resolvable RDF representation. This means that you can query it like a database and you can link to it in the data that you publish in as RDF - enriching both your own content and the semantic web as a whole. Cool.. Another project along similar lines but from the commercial side is called FreeBase.

Within the bioinformatics domain, we will shortly be seeing the public announcement of the new, fully RDFized version of UniProt and we already have an RDF wrapper around a large portion of the world's major bioinformatics databases in the form of Francois Belleau's Bio2RDF.

The semantic web is there - time to start writing those agents and browsers that can actually make use of it!

0 comments: