On Thu, 9 Sep 2010 14:30:52 +0100
Ian Davis<[email protected]> wrote:
This is based on the RDF conversion at
http://semanticweb.cs.vu.nl/lod/wn30/
How similar is your work to this version?
They're similar in that they're both based on Wordnet 3. There are some
key differences though:
1. The vu.nl version includes all of Wordnet. Mine is just the nouns.
2. The vu.nl version uses a SKOS-like modelling (though not SKOS):
"cat" for example has an rdf:type of something like "Noun". In mine,
"Cat" would have an rdf:type of rdfs:Class - i.e. I define each noun
as a class. I also include a parallel SKOS mapping of Wordnet 3 and
reference between them using rdfs:seeAlso/foaf:focus.
3. I define superclasses to group all reasonable interpretations of
English language words. e.g. a "Fool" superclass that is the union of
"Fool, as in incompetent person", "Fool, as in gullible person" and
"Fool, as in jester".
Very interesting! I'm curious though: what's the application scenario that made
you create this version?
And also, on this:
There's a class:
<http://ontologi.es/WordNet/class/Fool>
which acts as a superclass of all three senses of the word "fool". With
"fool", the meanings of each sense are close enough that the distinction
is not especially important, but with say "crack" this could refer to a
small gap (a doorway that's open a crack), a break (a crack in a mirror),
a sound (the crack of a whip) or crack cocaine - so distinguishing is more
useful.
How do you make the distinction between the two situations--I mean, based on
which elements in the Wordnet data?
Cheers,
Antoine
