Re: Heads Up - Important meeting Wednesday - PR Vote!!

I see some problems to decide next thursday to
request PR for the OWL Semantics document S&AS.

First, the text is not up to date with the RDF 
Semantics document.
It still contains the following lines:

"The specifics of datatype theories used here differ
slightly from those in the RDF semantics currently
under review.  It is expected that these discrepancies
will be resolved during last call, following which this
document will be revised to correspond directly to
RDF datatyping."

There are various discrepancies between S&AS
and the current RDF Semantics document.
For example, S&AS uses RDF's datatyped interpretations
for vocabularies instead of graphs, as is done in 
the current RDF Semantics editorial version.
(Actually, it seems that on this point it is not S&AS
but the RDF Semantics document that should change - I
sent a message about this to rdf-comments [1].)

The definitive text of the RDF Semantics document is not
yet available.  In fact, RDF Core are now discussing
the LC2 comments.  However, it is expected that the
definitive text is available soon.

In connection with the new RDF semantics several
technical, textual corrections need to be made to S&AS.
Wouldn't it be better to make these changes and review
these and then decide to go to PR with the document?

==

Since WebOnt's last call there have been many changes
to S&AS, and these are clearly visible in red in the
editor's draft.
It seems that it would be a good idea to go
further than checking whether the RDF-related
updates are correct, and to do a "mini review"
(before deciding about PR) of the main text, Sections 1 through 5, 
not to talk about water under the bridge but to focus on 
whether the document is up to date with RDF Semantics
and on whether the changes are not in conflict with each 
other.
Unfortunately, I have no time to do this before thursday.
Moreover, the final RDF part is not yet in so it seems better
to wait with such an action until that is in.

==

It should be noted that the appendix of S&AS that proves the 
correspondence theorem seems never been to have been reviewed in 
detail up to the point that a reviewer says yes, I consider 
this theorem proved.

I am concerned about the possibility of errors in the
correspondence theorem.
This is not improved by the remark of Ian during the last
telecon that "the correspondence theorem is broken".
Ian: perhaps you can give more concrete information about
this?

The situation is now that there are two normative descriptions
of OWL/DL entailment, one in Section 3 and one in Section 5.4.
Corresponding with this, there are two routes for description
logic reasoning with OWL in the form of RDF:
one that uses the definition in Section 5.4 and one that
uses the definition in Section 3 in combination with 
preprocessing and postprocessing from and to RDF.
The correspondence theorem claims that the two routes
are consistent.

In case they are not, I propose that Section 5.4 is made
informative.



Herman ter Horst

[1] http://lists.w3.org/Archives/Public/www-rdf-comments/2003OctDec/0148.html

Received on Wednesday, 12 November 2003 09:10:58 UTC