- From: Seth Russell <seth@robustai.net>
- Date: Wed, 19 Jun 2002 21:15:33 -0700
- To: "Thomas B. Passin" <tpassin@comcast.net>, <www-rdf-logic@w3.org>
From: "Thomas B. Passin" <tpassin@comcast.net> > Ah, back to quads. I'd like to be able to identify triples, too. In fact, > I remember thinking during the recent thread on "dark triples" (was that > here or on rdf-interest?) that the concept didn't seem workable to me > without being able to identify actual triples and to make statements about > them. Maybe I just didn't understand something crucial. Well it seems like there are at least two ways to do it: Method (1) - identify each triple ... give it an identify. Then you put triples in context by drawing a 'containedInContext' arc from it to it's context: <id1> <s1> <p1> <o1> <id2> <id1> <containedInContext> <c1> ... where the quad identified by <id2> puts the quad identified by <id1> in context <c1>. If it makes you feel squeemish when you say 'context' you can say 'graph' and feel better. Method (2) the context is a container of the triple. Here is the same statement as above: <c1> <s1> <p1> <o1> I like method (2), containment, because it seems like a natural progression: objects contained in predicates, contained in nodes, contained in contexts. Also you dont end up with a lot of wiggy <id> uri floating around to confuse the issue. In fact I liked it so well I wrote a demo in python to prove it works. Download and play with quads ... http://robustai.net/sailor/ Seth Russell
Received on Thursday, 20 June 2002 00:21:53 UTC