Re: LANG: syntactic version for imports (and other things)

From: Jim Hendler <hendler@cs.umd.edu>
Subject: Re: LANG: syntactic version for imports (and other things)
Date: Mon, 16 Sep 2002 22:55:00 -0400

> I now fully understand the difference in understanding between me and 
> Peter - turns out there is something in the DAML+OIL model theory 
> that I never realized -- according to separate info that I received 
> from Peter and others, the following turns out to be legal DAML 
> (syntactically)
> 
> 1)
> URI1 has an ontology called AAA, including class XXX
> 
> 2)
> URI2 has an ontology called BBB,
> and a class YYY that states it is a subclass of URI1:XXX.
> 
> if I understand correctly, however, I am told that this means that 
> while YYY makes this claim, since it has no imports statement to AAA, 
> it as if this subclass statement didn't exist.

I can't imagine how you came up with this idea.  The second ontology
certainly has all the effects of the subclass statement.  

What is true, at least as far as I can see, is that information about
referenced resources is not imported unless there is an imports statement.
That is, if document 1 contains a DAML+OIL ontology including
   YYY rdfs:subClassOf XXX .
and document 2 contains a DAML+OIL ontology including
   ZZZ rdfs:subClassOf YYY .
but no imports statements, then the ontology in document 2 knows nothing
about YYY being a subclass of XXX.

> I never realized this before, and think that if it is true, we should 
> fix it --currently a large number of the DAML documents out there 
> therefore do not mean what their authors intend -- we could tell them 
> all they are wrong because they don't follow the model theory, but 
> seems to me it might be better for us to make it possible for people 
> to have some meaning when they link to items in other documents 
> (instead of to whole ontologies) since they can do it anyway, and the 
> average user will assume the "natural semantics" as opposed to the 
> current situation (I put owl statements in my document but owl 
> reasoners simply ignore them - yecch).

As your understanding does not correspond with DAML+OIL there is nothing to
be fixed here.

> >  > One of
> >>  the problems we got into with DAML+OIL is we didn't say what imports
> >>  meant, and thus everybody interpreted it to mean whatever they felt
> >>  like.
> >
> >Well, I think that the problem was that we thought that it was obvious what
> >a DAML+OIL ontology was, and thus what imports meant.
> 
> no, I think several of us had different opinions, and we decided to 
> move to the Working Group process to work out a solution.   I think 
> we agreed what imports mean, but not what it meant to point to 
> something without importing!

There was discussion about DAML+OIL imports (mostly on www-rdf-logic).  As
you indicate, the main gist of it is whether reference without imports 
causes importing.  This was resolved, I strongly believe, in favour of
reference without imports not causing any importing.

[...]

peter

Received on Monday, 16 September 2002 23:39:13 UTC