Re: PROPOSAL to *close* (not postpone) ISSUE-83

This is why we have the task of drafting the wording (carefully :).

I haven't looked at the documentation carefully. I was trying  
specifically to address what is in the semantics document. My  
understanding is that in OWL Full, the fragment in question would  
have semantics because all pieces of syntax have semantics in OWL   
Full. However pieces of syntax were not given *extra* semantics if  
they weren't present in OWL-DL.

Issue-83 asked for support for the axiom. To the extent that this is  
construed to be a specific semantics in OWL-DL this is rejected. To  
the extent that this requires additional semantics specified for OWL- 
Full, this is not required, by the above reasoning, but is at the  
discretion of the editor/authors of the OWL Full semantics document  
(except to the extent that it might delay us).

To the extent that the user documentation might mention the meaning  
of the fragment in OWL Full, the only constraint would seem to be  
that the documentation accurately reflects the actual semantics.

At the meeting Jeremy was asked whether he thought it likely that  
such a fragment be supported by HP, he answered no. So deciding  
whether to include this kind of thing in the documentation might also  
take into account whether implementor support is likely to become  
available (so as to not set unrealistic expectations).

Anyways, that's the reasoning.

-Alan


On Dec 13, 2007, at 4:41 PM, Jim Hendler wrote:

> We don't document features outside DL in OWL 1.0?  The reference  
> document does exactly that - we didn't provide semantic discussion  
> of all possible features, but every vocabulary term in OWL was  
> discussed w/respect to DL and Full.   Grabbing one at random:
>
> ---- from Owl ref 4,1
> The property owl:equivalentClass is used to indicate that two  
> classes have precisely the same instances. Note that in OWL DL,  
> classes simply denote sets of individuals, and are not individuals  
> themselves. In OWL Full, however, we can use owl:sameAs between two  
> classes to indicate that they are identical in every way.
> ----
>
> so we did indeed document the differences.  There are things one  
> can do in OWL Full that we didn't document, but that was mainly  
> because we didn't think of them, not that we were not trying to  
> document them.
>
>
> On Dec 13, 2007, at 11:08 AM, Alan Ruttenberg wrote:
>
>>
>> This was discussed at TC. Action is on me an Ian to draft wording  
>> explaining closing/relation to OWL Full.
>> The gist is that it will never be in DL because of undecidability.  
>> It will be in OWL Full by virtue of syntax. Following precedent of  
>> OWL 1.0 we don't specifically document features outside OWL DL,  
>> though they may have semantics given in full, and that it will be  
>> at the discretion of the OWL Full semantics document editor to  
>> decide whether they want to provide a specific semantics for the  
>> form.
>> -Alan
>>
>> On Dec 13, 2007, at 10:05 AM, Peter F. Patel-Schneider wrote:
>>
>>>
>>> Issue-83 asks for property chains on both sides of subproperty  
>>> axioms.
>>>
>>> As pointed out by Uli Sattler
>>> http://lists.w3.org/Archives/Public/public-owl-wg/2007Nov/0600.html
>>> this makes OWL 1.1 undecidable.
>>>
>>> Contrary to what Ian Horrocks says
>>> http://lists.w3.org/Archives/Public/public-owl-wg/2007Dec/0012.html
>>> this feature would automatically be in an OWL
>>> Full version because there would have to be in OWL Full a semantic
>>> treatment of property chains and then there would be no way of  
>>> excluding
>>> them from both ends of a rdfs:subPropertyOf axiom.
>>>
>>> I therefore propose that we CLOSE ISSUE-83 without doing anything  
>>> on the
>>> twin grounds that it both compromises decidability in OWL 1.1 and  
>>> is not
>>> handled by tools, and that there is nothing special that needs to be
>>> done in OWL Full.
>>>
>>> Peter F. Patel-Schneider
>>> Bell Labs Research
>>>
>>> PS:  If the "undecidability" was not present above then it would be
>>>      reasonable to POSTPONE the issue.  However, undecidability
>>>      conflicts with the goals of OWL DL (and OWL 1.1) and thus I
>>>      strongly believe that CLOSURE is much more appropriate.
>>>
>>
>>
>
> "If we knew what we were doing, it wouldn't be called research,  
> would it?." - Albert Einstein
>
> Prof James Hendler				http://www.cs.rpi.edu/~hendler
> Tetherless World Constellation Chair
> Computer Science Dept
> Rensselaer Polytechnic Institute, Troy NY 12180
>
>
>
>