Re: owl:sameAs links from OpenCyc to WordNet

There seems to be a consensus that the RDF/OWL representation of WordNet 2.0
(as described at http://www.w3.org/TR/wordnet-rdf/ ) is not a representation
of the world at large but is rather a representation of a language-level
system that in turn represents the world at large.  In light of that, and
based on some of the suggestions that have surfaced in the course of this
thread, I've discussed some options with David Baxter here at Cycorp.  In my
mind, those discussions have led to the proposal I describe below.

In OpenCyc (in accord with Hugh's suggestion) include entities that
correspond directly with RDF WordNet Synset resources, and publish
owl:sameAs links between these entities and RDF WordNet resources, as
appropriate.   Also publish links from ordinary Cyc entities to the RDF
Wordnet resources using rdfs:seeAlso (in accord with suggestions from Fred
and Matthias).  Finally, publish links between the OpenCyc entities that
correspond directly with RDF WordNet Synset resources and the corresponding
RDF WordNet Synset resources with a new property defined in Cyc,
openCyc:synsetDenotes. "FOO openCyc:synsetDenotes BAR" will mean that FOO is
a synset that denotes the entity BAR.

 For example, for the case of  <
http://www.w3.org/2006/03/wn/wn20/instances/synset-India-noun-1>, we would
introduce an Cyc term to denote that object, <
http://sw.opencyc.org/2008/06/10/concept/en<http://sw.opencyc.org/2008/06/10/concept/en/India>
/wn-20-synset-India-noun-1<http://www.w3.org/2006/03/wn/wn20/instances/synset-India-noun-1>>.
We would then publish the following links:

(1)
"<http://sw.opencyc.org/2008/06/10/concept/en<http://sw.opencyc.org/2008/06/10/concept/en/India>
/wn-20-synset-India-noun-1<http://www.w3.org/2006/03/wn/wn20/instances/synset-India-noun-1>>
openCyc:synsetDenotes <http://sw.opencyc.org/2008/06/10/concept/en/India>"
(meaning that the NounSynset for the first sense form of "India" denotes
India)

(2)
"<http://sw.opencyc.org/2008/06/10/concept/en<http://sw.opencyc.org/2008/06/10/concept/en/India>
/wn-20-synset-India-noun-1<http://www.w3.org/2006/03/wn/wn20/instances/synset-India-noun-1>>
owl:sameAs <http://www.w3.org/2006/03/wn/wn20/instances/synset-India-noun-1
>".

(3)
"<http://sw.opencyc.org/2008/06/10/concept/en/India> rdfs:seeAlso <
http://www.w3.org/2006/03/wn/wn20/instances/synset-India-noun-1>".

By including rdfs:seeAlso links, we will accomodate the existing
technologies that seek out and leverage such links.

By including owl:sameAs links, we will accomodate the existing technologies
that seek out and leverage those links.

By including openCyc:synsetDenotes links, we will make it clear that the
synset entities in the WordNet ontology are at least quasi-linguistic things
which can mean or signify (denote) the "world at large" objects described by
the OpenCyc ontology.

I'd very much appreciate hearing about any red flags this proposal raises
for anyone.  I also thank all of you for your input in helping us to arrive
at what seems like a principled and workable approach.

Blake

Received on Wednesday, 25 February 2009 16:27:00 UTC