Re: LANG: need to CLOSE Issue 5.6 Imports as magic syntax

In short, we disagree, but that doesn't bind the
WG to either position.

To elaborate a bit...

On Wed, 2002-10-30 at 08:38, Jeff Heflin wrote:
> I believe we cannot in good conscience postpone imports. Until we
> resolve this issue, we cannot release a TR for the language.

I'm not sure what you mean by that; we certainly can resolve
the issue by postponing it, and release a TR without this
feature.

JimH points out the relevant requirement... I'm not so
sure we can't meet that without imports. To me, the use
of a property or class from another ontology is sufficient
" to explicitly extend other ontologies in order to reuse terms while
adding new classes and properties."

> In fact, I
> would strongly oppose releasing another working draft until we have a
> resolution on this issue.

I'm in an equally strong but opposed position; if the WG decides
to include this feature, I intend to have my objection recorded
for later review.

> Why is imports so important? Without it, we cannot build Semantic Web
> agent applications or have Semantic Web based e-commerce.

I build Semantic Web applications without imports all the time.
Some of them are e-commerce applications (though none
of them directly involve transfer of money).

I believe Mike Dean reports similar implementation experience...
yes...

"15:23:57 [lib-scrib]
- implementation experienc that don't need this stuff15:24:07
[lib-scrib]
MikeD: us too: we don;t need imports"
	-- http://www.w3.org/2002/10/07-webont-irc#T15-05-15

> Imports is the
> only feature in the language that guarantees people have common meaning
> for the same terms.

I don't see any guarantee. I see a technique that may or may
not help build confidence. My opinion is that it's counter-productive,
as it prematurely closes off other techniques.

> Any application where money or contracts is involved
> needs a definitive way of resolving what is meant that does not depend
> on a particular user or implementation policy decision. Imports is one
> way of doing this on the Semantic Web. Furthermore, imports has been a
> common feature in ontology languages for at least the last 10 years
> (although it goes by the name of ontology commitment, ontology
> inclusion, or ontology extension).

I'm familiar with this feature in SHOE; I'd appreciate pointers
to other ontology languages that have this feature.

> If we don't have this feature, then
> we don't have an ontology language and better change the names of our
> language and our working group.

I think we do.

> I have proposed a workable solution that is easy to specify

workable? Perhaps. easy to specify?
the proceedings of this WG seem to say otherwise.

> and that, as
> far as I know, does not break anybody's application.

I stipulate that it doesn't break any applications that
I can see.

But it does have a cost that is more than I'm not willing
to bear (premature standardization).

> I have
> implementation experience with this kind of solution (SHOE, one of the
> first fully-implemented Semantic Web languages has this as a central
> feature). I'm sure the others in the working group who have worked with
> ontologies have similar design experience.

I look forward to hearing from them.

So far, Mike Dean and I have reported that it's not needed.

I read Jos as saying he could go either way.


> It is critical that we resolve this issue as soon as possible.

I agree...

> I suggest
> we work hard to find a solution that everyone can live with and put it
> in our next working draft.

I have not seen a solution I can live with, and in the interest
of timeliness, I suggest we leave this for later.


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Wednesday, 30 October 2002 10:44:33 UTC