RE: ACTION: task force unasserted triples

> >
> > The key point of the request of the SWCG is to get precisely one
> > note/statement on the dark-triple requirement on which the people
> > involved all agree and which contains all the indicated elements
> > (motivation, example, etc.).
> >
> > Guus
>
> I believe that there have been several statments of the problem.  I'm
> willing to spend some time, once more, to put together something on the
> issue.   So I will at least add what Pat has said to what I have said.
> However, this has already passed the point of diminishing returns,
> and is getting very close to the point of negative returns.
>

I am rather more up-beat.

We have, IMO, made substantial progress towards clarity.

We have a number of proposals and a number of test cases.
It is possible to distinguish all the proposals by the entailments they
give on the test cases (no two proposals pass exactly the same test cases).

There are also some other methodological issues that need expression -
these express the difference between Peter and Pat who seem to dislike
using OWL to express OWL constraints, and Dan and myself who seem to like
such behaviour. I think this difference too can be reflected in test cases.

I think the DTTF can agree on a table of which test cases have which
results on which proposals (it is merely mathematics at that point).

Then, if we do a straw poll of the WG on the various test cases as to what
is the WG's preferred entailments we can motor forward.

The basis of the resolution at the f2f was that the appropriate entailments
require dark triples.
If we have consensus on three or four entailments and dispute on three or
four then we should be able to see if any non-dark triple proposals fit the
ones with consensus. If there are none, we are done, and we can go to RDF
Core saying this is waht we want to do, illustrated by these test cases. We
did consider some non-DT approaches but they didn't get these entailments
right. Hence DT are desirable.

If there are only non-DT approaches left standing after we consider the
WG's desires in terms of the test cases, we are also done.

If both DT and non-DT approaches support the desired entailments, then I
think we should indicate to RDF Core that we want the design freedom. That
is a very different sort of conversation to the one we agreed to try and
have after the f2f. However, Pat's message indicating how WebOnt can
explicitly use RDF as a carrier syntax and no more (at least in parts)
could be used as a way forward in this case.

Jeremy

Received on Thursday, 25 April 2002 10:02:51 UTC