Re: OWL Web Ontology Language

public-webont-comments@w3.org is probably not the best email list for this sort
of question, as the list is designed for comments about the OWL documents
themselves.  However, let me take a stab at your question anyway.  I've cc'ed
semantic-web@w3.org which is probably a better list for this kind of question.

From: "Bdn 01" <bdn_01@hotmail.com>
Subject: OWL Web Ontology Language
Date: 4 Mar 2006 14:59:33 -0800

> OWL Web Ontology (XML Presentation Syntax)
> Date: Sat, 04 Mar 2006 22:59:28 +0000
> Mime-Version: 1.0
> Content-Type: text/plain; format=flowed
> 
> Hello,
> 
> I have been following the development of the web ontology work for over a 
> year now and found the work on this topic extremely exciting. I do now 
> however have a concern that it does not cater for the situation where an 
> individual's property value may be neither a datatype, or another 
> Individual, but rather a class.
> 
> The reason I write is I would prefer to correctly apply the [OWL] standards 
> than extend the standards to cater for what (to me at least) seems to be a 
> standard case. I have posted the example below to help ilustrate this 
> scenario:
> 
> How to express that a child has a preference for dogs rather than cats as a 
> pet?
> - A given child is an individual with property "hasPetPreference".
> - A dog or a cat is a class (we should allow for identifying an actual beast 
> as an individual having class dog or cat).
> 
> Rgds
> Brendan


The problem with all such examples is determining just what they are supposed
to mean.  Do you want to say that the preference is somehow for the class "Dog"
versus the class "Cat", as, perhaps, one might prefer "PrimeNumber" to
"OddInteger" because the class of prime numbers are somehow more interesting
than the class of odd integer; do you want to say that the preference is
somehow for most dogs over most cats; or do you want to say that the preference
is somehow for all dogs over all cats.  True modelling should take the
differences into account.

All that said, there is an effort to extend OWL in a way that gives you a
light-weight way to at least state a relationship between an individual and a
class.  In this extension, OWL 1.1, class names can be also used as names of
individuals, so you could state

	Class(Dog)
	Class(Cat)
	ObjectProperty(hasPetPreference)
	Individual(john value(hasPetPreference Dog))
	Individual(jill value(hasPetPreference Cat))

For more information on OWL 1.1, see http://owl-workshop.man.ac.uk/OWL1_1.html

Peter F. Patel-Schneider
Bell Labs Research

Received on Thursday, 16 March 2006 19:25:53 UTC