Re: Paradox and Contexts

At 01:11 PM 12/2/00 +0100, Pierre-Antoine CHAMPIN wrote:
> > Note that the basic triple appears just once, even though it appears in two
> > quads.  This, I think, is roughly what happens when id= is applied to an
> > RDF property in its XML serialization.
> >
> > Actually, the approach I've taken for my contexts ideas is slightly
> > different:  the quad notation actually defines the reification, without
> > asserting the statement triple.  A statement is asserted in a context by an
> > RDF property that links the reification to the context.
>
>That is a question I had :
>which context does the "linking" statement belong to ?

A good question, indeed!

I think they may "belong to" any surrounding context in which the original 
context is held to apply.

If
   [C2] --asserts--> [C1]
then I think the 'asserts' statement in:
   [C1] --asserts--> [S1]
belongs to [C2].

In the contexts of McCarthy and Guha, there is a stack-like sequence of 
contexts associated with any statement.  In a logical sequence of 
operations, one can 'enter' a new context or 'leave' a context having the 
effect of pushing or popping the context stack:

Consider:

   (Enter C1)
   S1
   (Enter C2)
   S2
   (Leave)
   S3

Then the contexts associated with S1, S2 and S3 are:

   S1  (C1)
   S2  (C1,C2)
   S3  (C1)

My proposals use the 'asserts' property applied between contexts to achieve 
the effect of constructing a sequence of enclosing contexts (in a less 
general fashion than McCarthy/Guha) for a given statement.  This suggests 
to me that in the above sequence, the assertion ist(C2,S2) is a statement 
that "belongs to" C1:
   ist(C1,ist(C2,S2))



I don't have the relevant materials to hand as I type this, so I might be 
over-stretching the McCarthy/Guha idea of contexts.  Another answer to your 
question might be that the 'asserts' statements do not (necessarily) belong 
to any of the contexts referenced in the corresponding RDF graph, but 
belong to some implied context of discourse (a meta-context?).

#g

------------
Graham Klyne
(GK@ACM.ORG)

Received on Wednesday, 6 December 2000 10:08:29 UTC