Re: Fuzzy about what defines an ontologies' namespace

From: "Roger L. Costello" <costello@mitre.org>
Subject: Fuzzy about what defines an ontologies' namespace
Date: Wed, 26 Feb 2003 15:39:39 -0500

> 
> Hi Folks,
> 
> The OWL Guide is somewhat unclear about what defines an ontologies'
> namespace.  The Guide starts by saying that a default namespace should
> be declared:
> 
> <rdf:RDF xmlns="http://www.example.org/wine#"
> 
> The Guide says that this is used to "identify the namespace associated
> with this ontology".

This is unfortunate wording.  It should instead say something like
"The first namespace declaration provides a default namespace.  
Unprefixed element names are expanded into elements of this namespace."

There also needs to be some wording to the effect that it is conventional
to make the default namespace correspond to the Web location where the
ontology is located.

RDF is also currently wrestling with the differences between an XML
namespace, a vocabulary, and the Web location of an RDF document.  (I
actually don't think that they have yet put the third element into this
mix, but they should.)

> I took this to mean that, for example, the unique identifier for the
> Wine class is:
> 
>     http://www.example.org/wine#Wine
> 
> However, the Guide then followed up with a discussion of Ontology
> Headers.  In the example it shows this:
> 
>     <owl:Ontology rdf:about="http://www.example.org/wine"> 
> 
> It says this about the "about" attribute:
> 
> "The about attribute will normally be the URL of the current file. If
> desired, the ontology may be given a name that is a URN and independent
> of a particular physical location."
> 
> This lead me to think that perhaps the "about" attribute defines the
> namespace.  So, my question is this: is it the default namespace
> declaration which provides the unique identifier for each class and
> property defined in the ontology, or is it the "about" attribute?  

The name of an ontology could be totally different from the location of
ontology document.  However, really the only thing that matters for an
ontology is where it is located, as that is really the only thing that
matters for ontologies.  

The only thing that you can do with an ontology is load it (outside of the
OWL spec) or import it.  Loading would either be done directly by feeding
an ontology as a character sequence into a tool (which doesn't involve any
naming issues) or tell the tool to load the ontology document at a
particular Web location.  Importing is always done by looking for the
ontology document (if any) at a particular Web location.

> As a
> followup to that question: is xml:base not relevant with OWL?

It is, at least I think that it is.  This is, however, the realm of the RDF
specifications.  The namespace stuff is also the realm of the RDF
specifications.   

> Thanks!  /Roger

peter

Received on Friday, 28 February 2003 07:49:38 UTC