W3C home > Mailing lists > Public > public-owl-dev@w3.org > July to September 2007

RE: declaredAs

From: Michael Schneider <m_schnei@gmx.de>
Date: Tue, 14 Aug 2007 14:29:18 +0200
Cc: public-owl-dev@w3.org, phayes@ihmc.us
Message-ID: <20070814122918.231920@gmx.net>
To: davidt@hp.com

Hi, Dave!

David Turner wrote:

> Pat Hayes wrote:
>> Not really. Im sure it was meant to have this intention, but 
>> the effect of moving to punning is two-fold: it breaks the 
>> OWL Full semantics, and it breaks the semantic connection 
>> between OWL and RDF. 
> 
> I realise that this point has been elaborated on later in this thread,
> but I wanted to express my general agreement with it anyway. Punning is
> rather different from what OWL Full does, and you can observe this
> difference via owl:sameAs:
> 
> ex:A owl:disjointWith ex:B .
> ex:A owl:sameAs ex:B .
> _:x rdf:type ex:A .
> 
> is syntactically invalid in OWL-1.0 DL, inconsistent in OWL-1.0 Full and
> consistent in OWL-1.1 with punning (with thanks to Dave Reynolds)

Yes, this example is indeed trivially consistent in OWL-1.1, because, according to the punning idea, the URI 'ex:A' in the first and third line may denote a semantically completely unrelated resource than the URI 'ex:A' in the second line. In effect, these URIs have even to be regarded as being different URIs, which should better be called 'ex:A_Class' vs. 'ex:A_Individual', but are not for convenience, I suppose.

This is further explained in section 3.3 of

    http://owl-workshop.man.ac.uk/acceptedLong/submission_11.pdf
 
I like to call this a "kind of overloading". I already pointed out some months ago in 

    http://lists.w3.org/Archives/Public/public-owl-dev/2007JanMar/0008.html

around the time when I first saw this new feature, that I fear that punning will lead to hard to detect errors. I expect, people will thankfully use this feature as it gives them the /presumed/ freedom to talk, with the /presumably/ same URI, about the /presumably/ same resource, once as an individual, once as a class. But it will really turn out that both resources are semantically completely unrelated, what most users will probably not be aware of ("Hey, same URI, same resource, isn't it?"). And even for people who actually understand the punning "deception", I suppose it will not always be easy for them to detect in concrete ontologies the places where punning is applied.

My personal conclusion (well, at current :)): As long as I am not enforced to do so, I will most probably never use punning, when creating OWL-1.1 ontologies. And I would like to have a checkbox in my modeling tool, which, when opted (will always be opted in my case!), makes the tool warning me if punning is used somewhere within other people's ontology.

And this seems to build us again a little bridge back to the original topic of this thread, declarations, doesn't it? (oh dear!) ;-)


Cheers,
Michael

-- 
Michael Schneider <m_schnei@gmx.de>
Received on Tuesday, 14 August 2007 12:29:35 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:54 GMT