RE: partial response for LC comment 21 JDB2

>-----Original Message-----
>From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org]
>On Behalf Of Peter F. Patel-Schneider
>Sent: Thursday, February 19, 2009 10:27 PM
>To: public-owl-wg@w3.org
>Subject: partial response for LC comment 21 JDB2
>
>I have handled some of the comments in LC comment 21.  The other
>comments are for other documents.  Editors of these documents are
>welcome to augment the response.  I may also be able to make some of the
>changes.

I have added below text concerning the RDF-Based Semantics section (also put
into the Wiki).

There were 4 points, and I consider at least 3 of them mostly trivial and
moot now.

However, the second point about different datatype maps is in my opinion an
issue that needs discussion. For compatibility reasons, the notions of a
datatype and a datatype map in the RDF-Based Semantics are inherited from
the RDF Semantics spec ("D-entailment" = RDFS + datatype maps). In fact, the
definitions there are used at several important places in the OWL 2 Full
spec, and there are semantic conditions in D-entailment that need to be
supported by OWL 2 Full as well. 

I have extended these notions of datatypes and datatype maps from the RDF
Semantics conservatively in a way that they also support facets, and I have
put quite a lot of effort into aligning the facet-related stuff to that in
the Direct Semantics. This alignment has worked quite well so far. 

Nevertheless, this was merely an alignment of the main concepts (facet
space, and the like). Structurally, the two definitions of a datatype map
still differ significantly. For example, much of what is a part of a
datatype in RDF (for example the lexical space) is actually assigned to that
datatype by the datatype map in the Direct Semantics. In RDF, a datatype map
is simply a mapping (hence the name) from datatype IRIs to the actual
datatypes. In the Direct Semantics, a datatype map is a much more complex
object. In RDF, all of that complexity is in the datatype itself.

I just want to note that in OWL 1 DL, the definition of a datatype map was
pretty much the same as in RDF (at least structurally, of course without the
additional semantic conditions of D-entailment). This has changed notably in
OWL 2 DL, and I guess that Jos' comment targets this change. 

For the RDF-Based Semantics, I believe that I did all that what was possible
for me to have as much compatibility between the two notions of an "OWL 2
datatype map", without breaking compatibility with RDF. Conversion is easily
possible now (I even demonstrate this at one place in the document), but the
two notions are, after all, not the same.

Please read, with these considerations in mind, my proposed answers below.

Michael

>peter
>
>[Draft Response for LC Comment 21, partial]
>
>Dear Jos,

[...]

>=== RDF-Based Semantics ===

As a general note, please be aware of the fact that the RDF-Based Semantics
[1] is not yet a Last Call working draft, and it has been under considerable
edition since the last publication as of December. 

>- if there are differences between the datatypes and facets in tables
>2.2 and 2.3, on the one hand, and the datatypes and facets in that
>syntax specification, these should be mentioned.  If there are no
>differences, this fact should be mentioned as well.

It is indeed intended to have the same set of datatypes and facets in the
RDF-Based Semantics spec as in the Structural Specification. The working
group agrees that this should be more explicitly stated, since it does not
easily follow from the text in the published working draft. Therefore, the
working group plans to add clarifying text in the next published working
draft.

>- in my opinion, this document should use exactly the same definition of
>datatype map as the direct semantics specification.

First, note that the OWL 2 RDF-Based Semantics aims for full compatibility
with the semantics defined in the RDF Semantics specification [2]. The
semantics there already provides notions of datatypes and datatype maps, and
defines certain semantic conditions for them. In particular, as for OWL 1
Full, the central definition of an OWL 2 Full interpretation provided in the
RDF-Based Semantics document builds on top of the definition of a so called
"D-interpretation", as defined in the RDF Semantics specification, and by
this the existing definitions of datatypes and datatype maps from the RDF
Semantics specification are reused.

Further, since OWL 2 provides for the new concept of datatype facets, the
definition of a datatype, as given in the RDF Semantics specification, has
been extended by the notion of a facet space. Extending the definition of a
datatype is explicitly permitted by the RDF Semantics specification (see
section 5.1 of [2]).

Nevertheless, it is true that the different concepts used in the definition
of the extension for facets did not well match the concepts used in the
Direct Semantics specification in the last published working draft. This is
currently under revision, and the final outcome will be that the different
concepts are compatible with each other in that the different notions of
datatype maps can be easily transformed in each other.

>- above definition 4.2: "a OWL" => "an OWL"

Thank you for pointing us to this error, it will be fixed in the next
publication.

>- I do not understand the purpose of section 6.  There are no notions of
>consistency or entailment that could be implemented, so I don't see how
>anyone could exploit theorem 6.1.

The purpose of this section is to show how the two semantics of OWL 2, the
RDF-Based Semantics and the Direct Semantics, relate to each other. There
has been corresponding material in the OWL 1 specification [3]. Be informed
that at the time of the last publication, this section was in a very early
and incomplete state. A final and much enhanced version of this section is
planned for the next publication.

[1] OWL 2 RDF-Based Semantics, 2nd PWD: <...>
[2] RDF Semantics: <...>
[3] OWL Full (in S&AS): <...>

--
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

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