Re: WordNet in RDF/XML: 50,000+ RDF class vocabulary...

On Thu, 2 Dec 1999, Sergey Melnik wrote:
> Nice demo. Given the wealth of possible application scenarious, it'd be
> great to have an RDF dump of WordNet (a la Open Directory) that uses
> some "standard" RDF interpretation in a fixed namespace. Then people can
> start talking about the same "concepts".

I agree. As soon as we can give a Web identifier to the WordNet vocabulary
(or, specifically, the v1.6 version?) people can rush off and start using
it for Web annotations, content classification/categorisation,
accessibility tools etc. While any arbitrary URI would in principle allow
for this as long as folks agree, practicality and politeness both suggest
that the web identifier for WordNet is something that the Princeton
WordNet team should probably specify or bless.

So, that's one reason I copied wordnet@princeton.edu in on my msg. My
belief is that we know enough now to suggest how to deploy a
WordNet-in-RDF namespace, and that we could have a WordNet RDF vocabulary
(for nouns at least) up and running for semantic web applications fairly
quickly. 

I'm particularly interested to hear from anyone with contradictory views,
or with more experience of the WordNet vocabulary. WordNet proved easy to
mine for a set of RDF classes; I'm wondering whether there is a
mechanisable mapping from other parts of the WordNet to RDF
properties/relations too...

So - if anyone from wordnet@princeton is listening, you've been copied in
to a thread on the W3C RDF Interest Group [1]. The suggestion is that an
officially blessed WordNet 'namespace' URI (ie. a Web identifier such as a
URL or URN) would allow us to use WordNet concepts with the RDF [2]
information model for Web content classification and categorisation.

The URI could be something like http://cogsci.princeton.edu/rdf/wn/ or 
http://rdf.cogsci.princeton.edu/wn16/. Ideally those identifiers would
allow applications to connect to datadumps of WordNet using XML/RDF, but
the important thing for the use of WordNet with XML/RDF is that we agree
on a common identifier for the WordNet vocabulary. 

Dan

ps. I don't believe a separate RDF dump of wordnet is needed. Instead we
could use just a few line perl script to autoconvert the Prolog dumps of
WordNet that are already available. Similarly for an interactive version,
a short Perl script can wrap the commandline query interface.

[1] http://www.w3.org/RDF/Interest/
[2] http://www.w3.org/RDF/

Received on Thursday, 2 December 1999 23:22:08 UTC