RE: OWL dot OWL file

Hi!

A few things.

First, I did not add these label and comment triples to the RDF-Based
Semantics document, where, when used by an implementation, they would
clearly lead to unsoundness wrt to the RDF-Based Semantics. The additional
triples are only added to the owl2.owl proposal, where I added them for the
following reasons.

* owl.owl already /has/ a label per term, and my labels just follow the same
naming scheme (which is the same as in rdf.rdf and rdfs.rdfs). So not having
the labels would be a change, not the other way around.

* While owl.owl does not have comments itself, other namespace documents
have them, namely rdf.rdf and rdfs.rdfs. Why not simply aligning owl2.owl
with the other documents in this mainly editorial aspect?

Also, I don't see the point with the late stage here. owl2.owl won't be one
of our official deliverables, so its production should be pretty much out of
our normal schedule. I guess, if it is done just before the WG dissolves,
then it is fine. 

Let me say that I'm not even sure whether the production of owl2.owl is our
(the WG's) job at all. Were rdf.rdf and rdfs.rdfs be produced by the RDF WG,
or simply placed to their final positions by the W3C team afterwards?
Sandro?

The only thing that I want to make sure is that the really relevant
"semantic" triples (type, subclass, domain, range) of owl2.owl are backed by
one of our documents. Otherwise, owl2.owl would have no real relationship to
the OWL 2 standard in the end, which I would consider bad. So the set of
these triples have to be completed before we vote the documents into PR. But
this has nothing to do with additional labels and comments in the owl2.owl
file.

Finally, I don't want to see my original proposal as the final product yet.
It was meant as a basis for a discussion. There are at least several issues
that I would like to see resolved:

1) The original owl.owl had triples where some built-in classes being
subclasses of owl:Class instead of rdfs:Class and the like, and also the
domains and ranges of the built-in properties were often classes from OWL,
not RDFS. My current proposal deviates from this, basically only using
classes from the RDF(S) vocabulary, except for owl:Thing/owl:Nothing and the
top/bottom properties. This very much simplified things, but it may well be
seen as a wrong way to go by some. I could spend some work on finding out
what classes and properties should reuse the OWL classes instead, and this
would then also have an effect on my new list of axiomatic triples in the
RDF-Based Semantics. But I need a decision by the WG whether we should
either go the old owl.owl way (often referring to OWL classes) or the
simplified way (mainly referring to RDF(S) classes only).

2) The original owl.owl (and now the proposal) lists the four annotation
properties from RDFS (rdfs:label and friends). I would like to drop them
from the owl2.owl, since (1) they are not part of the OWL vocabulary, (2)
they are basically redundant (the terms are already covered by rdfs.rdfs,
which is even imported into owl.owl), (3) there is no precedence in rdf.rdf
and rdfs.rdfs that terms from other namespaces are reused, and (4) from a
"resolvable URI" point of view they would be invisible, since they have a
different base URI. In any case, a decision on this would have no
consequence for the RDF-Based Semantics, which *does* list these triples,
but other triples for terms of the RDFS vocabulary terms as well, which are
not mentioned in owl.owl.

3) What about the datatypes and facets. They are also mostly from other
namespaces (xsd), and owl.owl did not list the datatypes. So, I, personally,
would not like to see them in owl2.owl, but this should not be decided by me
alone. :-]

Please tell me what you think. And I think we should have this discussed on
our next telco.

Michael

>-----Original Message-----
>From: Ian Horrocks [mailto:ian.horrocks@comlab.ox.ac.uk]
>Sent: Sunday, July 26, 2009 11:46 AM
>To: Peter F. Patel-Schneider
>Cc: Michael Schneider; public-owl-wg@w3.org
>Subject: Re: OWL dot OWL file
>
>I agree with Peter. Given the very late stage of proceedings, can we
>please keep changes (w.r.t. owl.owl) to an absolute minimum.
>
>Ian
>
>
>On 24 Jul 2009, at 13:39, Peter F. Patel-Schneider wrote:
>
>> I have no major difficulties with such changes, except that every time
>> something is added - even an rdfs:comment - there is another chance to
>> have something go wrong.  To me, this argues very strongly for not
>> adding anything, and perhaps even going back to just including the
>> typing, domain, and range stuff.
>>
>> peter

--
Dipl.-Inform. Michael Schneider
Research Scientist, Dept. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: michael.schneider@fzi.de
WWW  : http://www.fzi.de/michael.schneider
=======================================================================
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, Az 14-0563.1, RP Karlsruhe
Vorstand: Prof. Dr.-Ing. Rüdiger Dillmann, Dipl. Wi.-Ing. Michael Flor,
Prof. Dr. Dr. h.c. Wolffried Stucky, Prof. Dr. Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
=======================================================================