Re: comment/question on OWL Reference

From: Benjamin Nowack <office@e-senses.de>
Date: Tue, 16 Sep 2003 15:26:24 +0200
To: public-webont-comments@w3.org

Benjamin,

Thanks very much for your comments. Sorry for the long delay in responding.

> Hi.
> 
> First of all I'd like to say that the various OWL 
> documents (and their authors) do a great job in 
> helping understand the underlying ontology concepts,
> including non-scientists like me. I enjoyed the 
> guide and could follow the reference quite well.
> But now some questions arose that I couldn't find
> an answer to in the reference:
> 
> The reference says that owl:Restriction is a subclass 
> of owl:Class. I understand that every 
> <owl:Restriction>
> ...
> </owl:Restriction>
> defines a separate class and must not have any 
> subnodes other than onProperty and the corresponding
> constraint. And usually a class is specified by not 
> only one restriction but many of them. Therefore
> it makes sense to combine property restrictions 
> via axioms such as owl:subClassOf.
> 
> But does that mean that it is not _allowed_ to 
> define a class like
> <owl:Restriction rdf:Id="MyClass">
> ...
> </owl:Restriction>
> 
> In 3.2.3 the Reference excludes owl:Restriction for
> being used as a named class but in 3.1 (NOTE 3) it
> is mentioned as a way to define complete classes 
> (if I get it right).

You're right. The note in 3.1 should have been more specific.
It now reads:

[[
NOTE: If one provides an RDF identifier for class descriptions of he 
type 2-6, this is not considered to be a class description, but a 
special kind of class axiom for complete classes.
]]

It should read:

[[
NOTE: If one provides an RDF identifier for class descriptions of
the enumeration, intersection, union or complement type,
this is not considered to be a class description, but a
special kind of class axiom for complete classes.
]]

This excludes named restrictions to be axioms for complete classes.
We will make this change in the upcoming editor's draft.

> 
> Another thing is that the Reference and the Guide
> give me the impression that it's best practice 
> to serialize a class with multiple property 
> restrictions by using multiple owl:subClassOf 
> properties (and not a single owl:intersectionOf 
> property instead). Is that true?

Multiple rdfs:subClassOf axioms have a different semantics that an 
owl:intersectionOf statement: necessary vs. sufficient conditions for 
class membership. For example, see the "Operetta" example at the end of 
section 3.2.1 with multiple subClassOf axioms. The accompanying text says:

[[
This class axiom states that an operetta is a musical work, that has a 
librettist and is not an opera. The use of the subclass relation leaves 
open the possibility that there are other musical works that have a 
librettist and are not operas. If we had wanted to say that operetta's 
are exactly those musical works that have a librettist but are not 
operas, we would need to use the owl:equivalentClass construct.
]]

Hope this clarifies the difference.

Thanks again for your comments,
Guus Schreiber


> 
> This may sound theoretical but it may have effects 
> on the way an OWL editor should best serialize its
> ontologies.
> 
> Thanks for your time,
> Benjamin
> 
> ___________________________
> benjamin nowack
> 
> am exerzierplatz 1
> D-97072 wuerzburg

-- 
Free University Amsterdam, Computer Science
De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands
Tel: +31 20 444 7739/7718
E-mail: schreiber@cs.vu.nl
Home page: http://www.cs.vu.nl/~guus/

Received on Monday, 17 November 2003 06:48:12 UTC