Re: Identifying OWL 1.1 files & the OWL namespace

>> 1) Do we assume that the additional OWL 1.1 vocabulary (such as
>> owl:SelfRestriction ;) ) will be added to the OWL meta vocabulary [1]?
>> Then the usual owl: prefix would be used, giving OWL files a more
>> consistent look.  However, I am not sure whether it is a recommended
>> practice to change a namespace that is now so widely deployed.
> 
> I'm quite sure it is a *bad* idea ;)

I think so too.

>> An
>> alternative would be to use a different namespace for the new 1.1 URIs -
>> this could help distinguish OWL 1.1 documents as well, but would also
>> mean that something like an owl11: prefix would be required.
> 
> a third option would be to put both old and new URIs in the OWL 1.1
> namespace, and use a unique prefix owl: corresponding to that new namespace.
> 
> - advantage: good-looking OWL files, and easy detection of the version
>   by looking at the namespace declaration
> 
> - drawback: OWL 1.0 agents would not recognize even the old URIs in OWL
>   1.1 files. But do we really want an OWL 1.1 file to be partially
>   parsed by a OWL 1.0 agent?

But we at least want this to work the other way around, so that OWL 1.1 
aware tools can process "legacy" OWL 1.0 files.  Furthermore, OWL 1.0 
tools can at least partially make sense of OWL 1.1 files.  Therefore I 
don't think that using a completely new namespace is a good solution.

The more I think about it, I would vote for a convention such as

   xmlns:owl="http://www.w3.org/2002/07/owl#"
   xmlns:owlx="http://www.owlx.org/2006/11/owlx#"  // or similar

Resulting files would contain statements such as

     <rdfs:subClassOf>
       <owlx:SelfRestriction>
         <owl:onProperty>
           <owl:ObjectProperty rdf:ID="likes"/>
         </owl:onProperty>
       </owlx:SelfRestriction>
     </rdfs:subClassOf>

This would allow tools and users to more easily recognize that OWL 1.1 
is used, and the owlx namespace declaration would point to the 
corresponding draft version.  For OWL 1.0 tools, the additional 
constructs might just become junk statements but at least they can 
process the vocabulary that they know about.

Any opinions from the draft editors?  The current RDF mapping indicates 
that the OWL 1.0 namespace would be used.  Why?

Thanks,
Holger

Received on Tuesday, 23 January 2007 17:35:39 UTC