W3C home > Mailing lists > Public > public-owl-wg@w3.org > February 2009

RE: draft response for LC comment 53

From: Michael Schneider <schneid@fzi.de>
Date: Mon, 16 Feb 2009 15:49:40 +0100
Message-ID: <0EF30CAA69519C4CB91D01481AEA06A001071533@judith.fzi.de>
To: "Ian Horrocks" <ian.horrocks@comlab.ox.ac.uk>, "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
Cc: "W3C OWL Working Group" <public-owl-wg@w3.org>
Ian Horrocks wrote:

>Sounds good to me.
>
>Michael: any comment?

Absolutely!

I have to say that I disagree with several points in this proposal. In
particular, the draft sais:

"""
It is the intention of the WG that in OWL 2 Full such importing
produces semantic inconsistency, although this is not yet in the OWL 2
Full Semantics document.
"""

I believe that this would be a very bad idea.

Why should Full provide such strong semantic support for a feature, for
which the Direct Semantics does not have any semantic treatment at all? We
normally do not want to have additional features in the Full semantics
compared to the Direct Semantics.
 
Why should there be semantic support for a feature, which is targeted to
ontology management in the first place, instead of logic? (So it should be
clear now that I would be against having this in the Direct Semantics,
either.)

Why should an implementer of an OWL 2 Full reasoner be bothered with this
all, or why should a user of such a reasoner accept the additional
complexity that results from such an additional semantic feature. This can
all equally well (and probably better) be done by an ontology management
tool on the syntax layer, just as for DL.

And note, the treatment of this "incompatibility" feature is actually only
marked as "SHOULD" in the Structural Spec. So I really see no necessity to
support this feature in the Full semantics. And the feature is realized via
an annotation property. A principle design idea in Full is to treat
annotation properties (and ontology properties) in the weakest possible way
(same idea as in the Direct Semantics, though there it's perfectly weak, of
course). Now this would be heavyweight semantics for an annotation property
instead.

I even put in question the idea of modeling the notion of incompatibility
via semantic inconsistency. I do not see what these two concepts have in
common. Two ontologies can be marked as incompatible for all kinds of
reason, the best one could say, perhaps, is that incompatibility will lead
to some sort of "unexpected behavior", whatever this means. But it does not
necessarily mean semantic inconsistency. 

And, finally, I would not even know how to appropriately define the
semantics for this feature, at least not without getting serious and
unexpected semantic side effects.
 
There is also a procedural point here. I do not see why the WG should make
such strong commitments to non-LC comments. I rather think that it suffices
to kindly thank the poster for his feedback, and say that the WG will take
his comments into consideration while continuing its work. 

Note that many of the points in this particular Non-LC comment are largely
outdated now, more than two months after the last publication, because I
have worked on this document since then. Specifically, I have completely
rewritten the section on "Ontologies" (this was long planned, I only did not
find the time so far). 

  <http://www.w3.org/2007/OWL/wiki/RDF-Based_Semantics#Ontologies>

The section is now basically restricted to saying that the syntax of OWL 2
Full is RDF graphs with IRI references. There is also an informative note
now on ontology management related content in ontologies, which is mainly a
pointer to Section 3 of the Structural Spec. In particular, I have removed
the original definition of an imports closure. This definition was
essentially taken from OWL 1 Full, where it played a role in the
correspondence theorem, but it isn't used anymore in the OWL 2 Full
document, so it became redundant.

So, after these changes, I guess the response needs to be redrafted, anyway.
Since this mainly becomes necessary due to my working on the RDF-Based
Semantics document, I am offering to take the task of redrafting the
response on myself, should you and Peter agree.

>Ian

Regards,
Michael

>On 13 Feb 2009, at 14:36, Peter F. Patel-Schneider wrote:
>
>>
>> [Response for LC Comment 53]
>>
>> Dear Jonathan,
>>
>> Thank you for your message
>>   http://lists.w3.org/Archives/Public/public-owl-comments/2009Feb/
>> 0000.html
>> on the OWL 2 Web Ontology Language last call drafts.
>>
>> The allowable ontologies OWL 2 DL form a syntactic subset of the OWL 2
>> Full ontologies.  For example, manipulation of the OWL built-in
>> vocabulary in certain ways is not allowable in OWL 2 DL but is
>> allowable
>> in OWL 2 Full.  Similarly, use of properties as both object properties
>> and data properties is not allowed in OWL 2 DL but is allowed in OWL 2
>> Full.
>>
>> Imports closures is another example where the syntax of OWL 2 DL is
>> less
>> permissive than the syntax of OWL 2 Full.  In OWL 2 DL, if
>> incompatible
>> ontologies are imported then the ontology is considered syntactically
>> invalid.  In OWL 2 Full, importing of incompatible ontologies is
>> syntactically valid (to allow any RDF as syntactically valid OWL 2
>> Full).  It is the intention of the WG that in OWL 2 Full such
>> importing
>> produces semantic inconsistency, although this is not yet in the OWL 2
>> Full Semantics document.
>>
>> This extends the situation in OWL 1, where owl:incompatibleWith had no
>> formal meaning, resulting in confusion as to exactly what it was
>> supposed to be used for.
>>
>> Although the two definitions of imports (in OWL 2: Syntax and OWL 2:
>> RDF-Based Semantics) are different in form, they are the same in
>> behaviour.  The WG will modify the wording in Section 2 of the OWL 2:
>> RDF-Based Semantics document, which is not yet at last call, to
>> conform
>> more closely to the wording in the OWL 2: Syntax document, and may
>> indeed make it point to the OWL 2: Syntax document.
>>
>>
>> Please acknowledge receipt of this email to
>> <mailto:public-owl-comments@w3.org> (replying to this email should
>> suffice). In your acknowledgment please let us know whether or not you
>> are satisfied with the working group's response to your comment.
>>
>> Regards,
>> Peter F. Patel-Schneider
>> on behalf of the W3C OWL Working Group
>>

--
Dipl.-Inform. Michael Schneider
Research Scientist, Dept. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: schneid@fzi.de
WWW  : http://www.fzi.de/ipe/eng/mitarbeiter.php?id=555

============================================================================
==

FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts
Stiftung Az: 14-0563.1 Regierungspräsidium Karlsruhe
Vorstand: Prof. Dr.-Ing. Rüdiger Dillmann, Dipl. Wi.-Ing. Michael Flor,
Prof. Dr. rer. nat. Dr. h.c. Wolffried Stucky, Prof. Dr. rer. nat. Rudi
Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus

============================================================================
==



Received on Monday, 16 February 2009 14:50:24 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 16 February 2009 14:50:24 GMT