Re: comment on O

Many examles in OWL Test cases have been validated by at least three 
different implementors as being in OWL Lite or OWL DL without such a triple.
 From the implementors' point of view, not fixing this in S&AS would be the 
change.

For forms sake:

I propose that OWL Lite and OWL DL ontologys should not require any triple 
that explicitly uses owl:Ontology.


Jeremy



Jim Hendler wrote:

> At 11:37 AM -0400 9/16/03, Peter F. Patel-Schneider wrote:
> 
>> From: Jeremy Carroll <jjc@hpl.hp.com>
>> Subject: comment on O
>> Date: Tue, 16 Sep 2003 16:57:14 +0200
>>
>>>  http://www.w3.org/TR/2003/CR-owl-semantics-20030818/mapping.html#4.1
>>>
>>>  The second rule seems to *require* that every ontology includes a 
>>> triple
>>>
>>>  xxx rdf:type owl:Ontology .
>>>
>>>  where xxx is either the name of the ontology or a blank node.
>>>
>>>  This does not appear to be the intent elsewhere.
>>
>>
>> Hmm.  Where elsewhere?
>>
>>>  The triple
>>>  O rdf:type owl:Ontology .
>>>
>>>  could be included in the Annotation rules instead.
>>
>>
>> Yes, this would fix a potential problem if the above is optional with
>> annotations on anonymous ontologies.
>>
>>>  Jeremy
>>
>>
>> I am actually in favour of making the type triple optional in this
>> production.  I think that it would require a working group decision at 
>> this
>> point, however.
>>
>> peter
> 
> 
> I would be happy to see it made optional, my second choice would be to 
> document the heck out of this -- it is now becoming so easy to 
> accidently put things in OWL Full because of things like this (or like 
> putting a max and min cardinality in the same restriction) -- I think 
> the main "fix" is to make sure that where there are things like this, 
> and we see them happen multiple times, we try to add some words to Guide 
> and/or Ref -- this makes the changes purely editorial without changing 
> our design
>

Received on Wednesday, 17 September 2003 11:31:29 UTC