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

Annotation entailment!

From: Denny Vrandecic <dvr@aifb.uni-karlsruhe.de>
Date: Thu, 28 Jun 2007 23:43:09 +0200
Message-ID: <20070628234309.okbdw3msgnsc08c8@webmail.uni-karlsruhe.de>
To: Alan Ruttenberg <alanruttenberg@gmail.com>
Cc: OWL list <public-owl-dev@w3.org>

OK, I checked the specs, and Alan is correct. OK, I can't derive it 
from the specs (this would require some in-depth reading), but actually 
there are normative test cases that answer the question pretty clearly 
(referenced below).

So, owl:sameAs does indeed carry annotation property entailments, i.e.

ex:A rdf:label "Grue".
ex:A owl:sameAs ex:B.
--------------------
ex:B rdf:label "Grue".

*does* hold. Furthermore,

ex:A rdf:label "Grue".
ex:A owl:equivalentClass ex:B.
----------------------------
ex:B rdf:label "Grue".

does *not* hold, but

ex:A rdf:type owl:Class.
ex:B rdf:type owl:Class.
ex:A rdf:label "Grue".
ex:A owl:sameAs ex:B.
-----------------------
ex:B rdf:label "Grue".

*does* indeed hold, but only in OWL Full (it's not a valid OWL DL ontology).

I guess

ex:A rdf:label "Grue".
ex:A owl:equivalentProperty ex:B.
----------------------------
ex:B rdf:label "Grue".

does not hold as well.

I find the fact that sameAs and equivalentClass are treated differently 
with this respect unintuitive, but at least it is well specified :) A 
well defined spec is the next best thing to an intuitive spec.

Thanks for your help,
denny

Here are the test cases I derived this from:
<http://www.w3.org/TR/owl-test/byFunction#function-sameAs>
<http://www.w3.org/TR/owl-test/byFunction#function-equivalentClass>

Zitat von Alan Ruttenberg <alanruttenberg@gmail.com>:

>
> I have recently understood it to be the case that this *is* entailed. 
>  However, I don't think that it is desirable, and it seems to be the  
> case that there is no OWL-DL reasoner that attempts to be sound and  
> (otherwise) complete that implements this behavior, at least that I'm 
>  aware of.
>
> What do you think of this behavior? Is it is what you expected or  
> wanted? If so, I'd be interested hearing about your use case.
>
> Until recently it was my (flawed) understanding that statements  
> involving annotation properties were to be ignored by a reasoner.
>
> -Alan
>
>
> On Jun 27, 2007, at 6:06 PM, Denny Vrandecic wrote:
>
>>
>> Does
>> ex:A rdf:label "Groo".
>> ex:A owl:sameAs ex:B.
>> entail
>> ex:B rdf:label "Groo"?
>>
>> i.e. are annotation property instances connected to the URI or the  
>> underlying individual?
>> (And respectively for classes and properties)
>>
>> Wondering,
>> denny
>>
>>
>
>
>
Received on Thursday, 28 June 2007 21:43:20 GMT

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