W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > October 2001

Re: Active issue rdfms-graph; formal description of properties ofan RDF graph

From: Graham Klyne <Graham.Klyne@MIMEsweeper.com>
Date: Sun, 21 Oct 2001 11:10:33 +0100
Message-Id: <5.1.0.14.2.20011021104556.00a0b7b0@joy.songbird.com>
To: Dan Connolly <connolly@w3.org>
Cc: Pat Hayes <phayes@ai.uwf.edu>, w3c-rdfcore-wg@w3.org, pfps@research.bell-labs.com
At 09:43 AM 10/20/01 -0500, Dan Connolly wrote:
>And no, I don't remember how much trouble we had with
>this at the F2F; I wasn't there the 2nd day. The
>previous model theory, which was based on n-triples
>concrete syntax, worked just fine for me.
>
>Let me review the record of the 2nd day...
>
>It seems to come down to this:
>
>   Pat: the realization that I have, if I do the MT as attached
>   to the graph, then issues like scope of exist quant
>   go away becauset there are no scopes in the graph.
>
>   --
>http://www.w3.org/2001/sw/RDFCore/20010801-f2f/2001-08-02.html#T16-15-50
>
>If you substitute "set of triples..." as above
>for graph, the epiphany holds, no?

Not quite.  With a graph, each node *is* its own identity;  there is only 
one (abstract) syntactic occurrence of each distinct node, so there's no 
need to define a scope -- or, to put it another way, giving every node 
universal scope works just fine, and document boundaries don't have any 
bearing on the interpretation of any given set of statements.

But when a serial (1-dimensional, sequence) syntax is introduced, such as 
N-triples, it is necessary to use labels in a way that multiple syntactic 
occurrences can be understood to refer to the same thing.  Specifically, 
the node identifiers used for this purpose in N-triples (formerly known as 
bNodes) only work if they are clearly scoped;  multiple occurrences of _:a 
in one document refer to the same thing, but not the same thing as _:a in a 
different document.  The trouble is, we don't seem to have a clear, formal 
idea of what constitutes an N-triples document.  And if/when we do have 
such an idea, we also need to have words that describe what happens to 
these node identifiers when documents are combined.  All of this is 
reasonably solvable, but with some increase in specification complexity and 
(at the time) going with abstract graph syntax avoided having to tackle 
this complexity.

In saying this, I'm not arguing against your proposal, just trying to 
explain the nature of the epiphany.

When I agreed to your proposal
   http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Oct/0186.html
I was mainly agreeing with the bit that the model theory should not be 
constrained by the current RDF syntax in what it describes:
   http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Oct/0193.html
rather then specifically agreeing to your abstract syntax proposal.

My current feeling is that if we can avoid the scoping issues then hanging 
interpretations on the abstract graph syntax makes the formalism 
easier.  But I'm beginning to think that, one way or another, the scoping 
issue may reappear in another guise in which case I would see your proposed 
abstract syntax as good as any.

#g


------------------------------------------------------------
Graham Klyne                    MIMEsweeper Group
Strategic Research              <http://www.mimesweeper.com>
<Graham.Klyne@MIMEsweeper.com>
------------------------------------------------------------
Received on Sunday, 21 October 2001 08:15:32 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:41:08 EDT