W3C home > Mailing lists > Public > public-swbp-wg@w3.org > January to March 2004

Re: WordNet Task Force - work outline

From: Aldo Gangemi <a.gangemi@istc.cnr.it>
Date: Wed, 31 Mar 2004 08:56:20 +0000
To: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Cc: public-swbp-wg@w3.org
Message-id: <1080723380.406a87b472569@imp.rm.cnr.it>

>At 16:35 +0100 30-03-2004, Jeremy Carroll wrote:
>I still owe an example of a simple use of WordNet ...

>this wasn't quite the one I had in mind, but has the advantage of being more 
>real:

>http://norman.walsh.name/knows/where#dk-copenhagen

>is described as having RDF type

>http://xmlns.com/wordnet/1.6/City

>clicking on the above URL gives an RDF/XML download ....
>and uses words from WordNet as RDF classes ...

>I think what Norm is trying to do is simply say that his copenhagen URI is a 
>resource which belongs to a class with some (strong) relationship to the 
>english word city in wordnet 1.6 sense 1 with description 'a large and densely 
>populated urban area; may include several independent administrative 
>districts; "Ancient Troy was a great city"'

>This use of WordNet is fairly naive (in the sense that non-experts can do it) 
>and does not depend on any of the relationships between words.

>Jeremy

I agree, indeed this is what I expect from Wordnet on the Semantic Web: the 
Copenhagen Norm is talking about is an instance of 
http://xmlns.com/wordnet/1.6/City. All right.

But also look at the file at http://xmlns.com/wordnet/1.6/City: City is a class 
introduced with all its taxonomic branch (poor practice: if each class is 
introduced with all its superclasses, the ontology results unnecessary long), 
then all hyponyms of "City" are introduced, for instance:

 <Class rdf:about="http://xmlns.com/wordnet/1.6/Dunkerque">
  <subClassOf rdf:resource="http://xmlns.com/wordnet/1.6/City"/>
    <label>Dunkerque</label>
 <comment>a city in northern France on the North Sea where in World War II 
(1940) 330,000 Allied troops had to be evacuated from the beaches at Dunkirk in 
a desperate retreat under enemy fire</comment>
  </Class>

But Dunkerque can be hardly a class, except in some peculiar ontologies that 
take 4D and set-theoretically extensional classes (Dunkerque as the temporal 
worm equivalent to the set of all the states in the area named Dunkerque!).

In fact, the hyperonym relation in WordNet *generally* means subClassOf, but 
words and terms often name individuals, and many individuals are so important 
to be put in a dictionary as Wordnet. Cities are a case.

Then, when reengineering WordNet for the SW, WN data types should be remapped 
to OWL (or RDF) data types in a consistent way, and I think our job is to 
explain as much as possible how to obtain that consistency.

Another curious thing: I supposed Copenhagen being there, but it is not. I 
checked WordNet, and the reason is that Copenhagen is a hyponym of "Capital 
City", which is a hyponym of "City", then it is not included in the resource.

Moreover, putting both "Capital City" and "Dunkerque" as hyponyms of "City" is 
patently a poor practice of ontological modelling (but not necessarily for a 
dictionary such as WN).

Thanks for pointing at that spontaneous example. The use of Norm is not na´ve, 
it is precise. What is too simplistic is the type mapping encoded in the RDFS 
file.

Cheers
Aldo



-- 
Aldo Gangemi
Research Scientist
Laboratory for Applied Ontology
ISTC-CNR
Via Nomentana 56, Rome, Italy
+39.06.86090249

-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/
Received on Wednesday, 31 March 2004 04:22:30 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 31 March 2004 04:22:52 EST