RE: reification/annotations

IIRC, you (Alan) and I brought up this discussion, but it has never been
discussed any further by anyone else.

Let me say that I would be surprised, if OWL editors such as Topbraid
Composer would really drop the original axiom triple, when the axiom is
annotated. This would make things more complicated for such tools (let alone
for people who want to have a look at the produced RDF code), without
bringing any good in exchange (except for the minor advantage a triple less
in the serialization). 

And keeping the axiom triples would not be against the spec, because I don't
think that we can disallow to have the axiom triple in the ontology, too. In
particular, the original axiom might be part of an imported ontology, to
which I have no access at all. So I would have to create the reified
annotation *in addition* to the axiom triple in some cases, anyway.

On the other hand, having both the axiom triple and the reified annotation
brings the advantage of allowing a clear separation between the actual
axioms and the annotations. One can then, for example, have different files,
where the annotation document imports the axiom document. So people, who
don't want to care about the annotations, can easily ignore them. This may
become relevant if there are masses of annotations for some ontology. I
would, myself, be in favor of such a "stripping" practice, at least
occasionally.

A minor addition: Having the original triple in the ontology would not make
it necessary for OWL 2 Full to have a special semantic condition, which
reconstructs the original axiom from the reified statement. 

Michael

>-----Original Message-----
>From: public-owl-wg-request@w3.org [mailto:public-owl-wg-request@w3.org]
>On Behalf Of Alan Ruttenberg
>Sent: Wednesday, May 28, 2008 6:00 AM
>To: Peter F. Patel-Schneider
>Cc: public-owl-wg@w3.org
>Subject: Re: reification/annotations
>
>I'm recalling that we discussed it and thought we had resolved it but
>was not sure, hence the question. If someone can pull out the discussion
>where (if anywhere) we decided this one way or another, then that would
>be helpful. Otherwise we can discuss it tomorrow.
>
>(Tracker: this is relevant to Issue-67 and Issue-81)
>
>
>-Alan
>
>On May 27, 2008, at 11:34 PM, Peter F. Patel-Schneider wrote:
>
>
>		Had we not decided to include the triple being reified in
>the
>
>		serialization when using reification?
>
>
>
>
>	My recollection is that indeed we have not decided to include the
>triple
>
>	being reified in the serialization when using reification.  (Of
>course,
>
>	my recollection may be faulty, but then where is the discussion
>and
>
>	resolution for the change?)
>