- From: Graham Klyne <gk@ninebynine.org>
- Date: Fri, 19 Dec 2003 17:56:19 +0000
- To: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>, "Chris Bizer" <chris@bizer.de>, <www-rdf-interest@w3.org>
At 13:31 19/12/03 +0100, Jeremy Carroll wrote:
>I suggest that instead of a quadruples or quintuples approach that this
>difference in semantics is better reflected by naming graphs (sets of
>triples).
FWIW, this is just what I do in my current software.
But I note that it is, in information terms, pretty much equivalent to
using quads, where the 4th item is a context id.
quads->labelled graphs:
collect statements by context id
use context id as graph label
put statements, without context id, into appropriately labelled graph
labelled graphs->quads:
for each statement of each labelled graph:
construct a quad that is the statement
PLUS the graph identifier as its context id
It's informally stated, but I think that's a bijection.
But, then you say:
>whose to say that there is not another quad somewhere
I think you have a point here, but I'd also ask:
Who's to say that one of your named graphs doesn't have another triple
somewhere?
Or, to put it another way, what do you do with:
ID => { < a, b, c >
< a1, b1, c1 >
< a2, b2, c2 > }
ID => { < a3, b3, c3 > }
?
#g
--
BTW, Chris, in my original response to you, I didn't advocate using
reification (though I did once argue its use as a way to abstractly
represent quads in "pure" RDF).
------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact
Received on Saturday, 20 December 2003 05:16:25 UTC