Re: question about imports

From: pat hayes <phayes@ai.uwf.edu>
Subject: Re: question about imports
Date: Thu, 5 Dec 2002 19:28:41 -0600

> >From: pat hayes <phayes@ai.uwf.edu>
> >Subject: Re: question about imports
> >Date: Thu, 5 Dec 2002 16:54:24 -0600
> >
> >>  >From: pat hayes <phayes@ai.uwf.edu>
> >>  >Subject: question about imports
> >>  >Date: Thu, 5 Dec 2002 12:27:45 -0600
> >>  >
> >>  >>
> >>  >>  As I understand things, the current meaning of imports is that including
> >>  >>
> >>  >>  owl:imports B
> >>  >>
> >>  >>  in a document A has exactly the same meaning as copying the imports
> >>  >>  closure of B into A.
> >>  >>
> >>  >>  So I have a question: consider two documents A and AC which are
> >>  >>  identical except that A contains owl:imports B, and AC actually has
> >>  >>  the imports closure of B copied into it at that point, but has no
> >>  >>  reference whatever to B.
> >>  >
> >>  >This is different from the previous proposal in that it drops the imports
> >>  >triple.
> >>  >
> >>  >>  These two documents have exactly the same
> >>  >>  meaning, right?
> >>  >
> >>  >No, because the second is missing a triple.
> >>
> >>  They are syntactically distinct, but I believe they are true in
> >>  exactly the same interpretations (?). Keeping the imports triple with
> >>  the imported graph is like saying P and P instead of P, right?
> >
> >Huh?  Are you thinking that this is a dark triple?
> 
> No, of course not. But the truth-conditions on owl:imports B are 
> exactly the same as the truth conditions on the imports closure of B, 
> right (??) If not, what *are* the truth conditions on owl:imports?

Close, but there is no way that OWL can turn off the RDF interpretation of
the triple, is there?  The imports closure conditions can only add to this
base condition, at least as far as I can figure.

> >  When did these come back?
> >
> >>  >>  And the first, but not the second, refers to another
> >>  >>  document.
> >>  >
> >>  >And they have different meaning (as n-triples documents).
> >>  >
> >>
> >>  In what does the difference reside?
> >
> >I thought that every triple made an assertion.
> 
> I never said otherwise. That is beside the point of my question. What 
> assertion does the imports triple  (considered in isolation, a single 
> triple) actually make, other than that imports closure be true? 

Just the same as any other triple, no?

> (And 
> BTW, what is the subject of that triple, and how does it enter into 
> the truth-conditions?)

Well, this is one of the problems of using triples.  The current proposal
is that the subject of the triple is unimportant and is (conventially)
written as "".

> >If this is no longer true,
> >then I'm going to have a pile of changes to make.
> 
> Not only is it true, one can say more: it asserts that the <s,o> pair 
> is in the extension of the property. What are the semantic conditions 
> on IEXT(I(owl:imports)) ?

Well, the editor's draft of the OWL semantics treats owl:imports specially.
For details see the draft.

> Pat


peter

Received on Thursday, 5 December 2002 21:55:16 UTC