Re: Treating a class as both an individual and a class?

From: "Roger L. Costello" <costello@mitre.org>
Subject: Re: Treating a class as both an individual and a class?
Date: Mon, 10 Mar 2003 07:42:57 -0500

> 
> "Peter F. Patel-Schneider" wrote:
> > 
> > > Suppose that I want to use this River class as an individual, as 
> > > shown here:
> > >
> > >     <Country rdf:ID="China">
> > >         <hasFeature rdf:resource="http://geodesy.org#River"/>
> > >     </Country>
> > >
> > > Note that the property "hasFeature" has as its value the class 
> > > River. Thus, the River class is being treated as an "individual".
> > 
> > Hmm.  I don't think that this is correct modelling.  You are *not* 
> > saying anything related to whether China has a river in it. 
> 
> What I was trying to express in this example is that the country China
> possesses the geographic feature "River".  I agree that it is a poor
> example.  Let me give another example. 
> 
> In this example, I would like to express some of the major features of
> my hometown of Davenport, Iowa:
> 
>     <City rdf:ID="Davenport">
>         <hasFeature rdf:resource="http://geodesy.org#River"/>
>         <hasFeature rdf:resource="http://geodesy.org#Interstate"/>
>         <hasFeature rdf:resource="http://geodesy.org#Airport"/>
>     </City>
> 
> Is this good modelling?  Does it effectively model the major features of
> my hometown?  Note that River, Interstate, and Airport are "classes". 
> Thus, in this example I am treating these classes as individuals.

No, this is not good modelling.  If you want to say that Davenport has a
River, you should say so, via

     <City rdf:ID="Davenport">
         <rdfs:subClassOf>
	    <owl:Restriction>
		<owl:onProperty rdf:resource="hasFeature" />
		<owl:someValuesFrom rdf:resource="http://geodesy.org#River"/>
	    </owl:Restriction>
         </rdfs:subClassOf>
         ....
     </City>


> > However, you do need to remember that metaclasses may not give you all 
> > that you might think that they are giving you.
> 
> A very interesting statement Peter.  Would you elaborate please?  /Roger

Many uses of metaclasses require more-powerful mechanisms than the ones
provided by RDF, RDFS, or OWL to capture their meaning.  Using metaclasses
in these cases generally just deludes the user into thinking that more is
going on than really is.

For example, think of what good owl:TransitivePropery would be without the
OWL model theory to back it up.  Then think of just what you would get by
using a similar name all by itself.

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Monday, 10 March 2003 07:54:24 UTC