Re: [WNET] Schemas online

Hi Ralph,

>> Of course the same effect could be reached by inserting that OWL Datatype/ObjectProperty are subClassOf rdf:Property,
> 
> And that's already declared in the OWL ontology [1,2] so we should
> not need to repeat the statement.

Just a note: someone who loads WordNet with RDF(S) infrastructure won't 
probably (want to) load the OWL schema, so s/he either needs the 
rdf:type statements I've added or the owl:Datatype/ObjectProperty 
subClassOf rdf:Property statement.

>>>> - I used XML Schema Datatypes also for e.g. the gloss and lexicalForm properties, but I can recall that when properties have an XSD as range then the actual glosses and lexical forms are not supplied with the XML lang attribute. If this is correct, I should probably change to rdfs:Literal?
> 
> This (how language tagging is identified) is a sore point for me.
> In general I would like to see non-English rdfs:label values, etc.
> but I thought that you'd decided that the Princeton WordNet 2.0
> data was exclusively English-language so we don't really need
> to carry explicit language tagging?

We know that Princeton data is EN-US, but why not also giving that 
information to the infrastructure? When you're mixing triples from 
different sources (e.g. different WordNets) the xml:lang is probably 
easier than keeping a list of which sources are in which language in 
your application.

> However, it's also not obvious to me what benefit we obtain
> from restricting the range of wn:lexicalForm and wn:gloss to be
> xsd:string rather than rdfs:Literal.  Unless it's clear that there
> is some utility to restricting the range my default is to permit
> the broadest class.  Let's not forget, too, that the Princeton

Good argument. I will change to rdfs:Literal.

>>>> - there are two properties that are subPropertyOf rdfs:label / rdfs:comment. The OWL validator complains that the rdfs properties should be DatatypeProperties. Is this an issue?
> 
> Not for me :)  If declaring wn:lexicalLabel and wn:hyponymOf
> to be rdfs:type owl:DataTypeProperty helps, that may be OK.

_lexicalForm_ and _gloss_ are already owl:DatatypeProperty. The problem 
  is, I think, that in OWL DL the DatatypeProperties and 
AnnotationProperties should be disjoint. (Of course in OWL Full anything 
goes.)

The rdfs:label and rdfs:comment are owl:AnnotationProperty in OWL DL. 
Changing wn:gloss and wn:lexicalForm to be owl:AnnotationProperty is not 
a solution, because AnnotationProperties are not allowed to have 
subproperties [1].

The reason I still want to keep gloss and lexicalForm subproperties of 
rdfs:comment and rdfs:label is that much RDF(S) infrastructure is 
programmed to make use of rdfs:label/comment in displays of the data.	

Therefore I would think the best course of action is to leave things the 
way they are. Actual users will probably not have much use of OWL DL 
anyway in the case of WordNet (e.g. we, as developers, should already 
make sure the instances are correctly classified). Comments on this very 
welcome.

Mark.

[1]http://www.w3.org/TR/2004/REC-owl-ref-20040210/#Annotations

-- 
  Mark F.J. van Assem - Vrije Universiteit Amsterdam
        markREMOVE@cs.vu.nl - http://www.cs.vu.nl/~mark

Received on Tuesday, 16 May 2006 09:03:59 UTC