Re: rdf inclusion

   [Dan Brickley]
   FWIW, I've rigged the namespace http://xmlns.com/wordnet/1.6/ to return
   chunks of the wordnet noun hierarchy projected into RDF classes.
   deferencing http://xmlns.com/wordnet/1.6/Tree or .../Car or .../Person
   gets a fragment of the larger (much larger) ontology. Don't look too
   closely, the generated markup has bugs, but the scenario should at least
   be reasonably clear.

It gets the superclasses and subclasses of Tree, right?  Starting with
Yellowwood, and ending with "Tree_of_Knowledge" (a rare species not
seen since Genesis).

   I don't claim this provides a 'natural definition', but it does seem more
   useful than having the entire multi-megabyte wordnet RDF thing be
   downloadable at the schema namespace URI.

I am not too familiar with Wordnet, but it doesn't seem to provide
much more than a subclass hierarchy + synonym links (?) + comments.  I
agree that there is a natural definition of "subgraph pointed to" for
a subclass hierarchy (namely, the subclasses and superclasses of the
nodes you start with).  As soon as you put anything else in, the
definition breaks down.  If someone wants to use "#Dog" and "#Tree"
from an ontology, they want to know more than, say, are there more
breeds of dog or species of tree (as far as this system believes)?

                                             -- Drew McDermott

Received on Friday, 24 May 2002 07:26:37 UTC