- From: Dan Brickley <Daniel.Brickley@bristol.ac.uk>
- Date: Tue, 21 Dec 1999 13:03:42 +0000 (GMT)
- To: Daniel Veillard <Daniel.Veillard@w3.org>
- cc: www-rdf-interest@w3.org
On Tue, 21 Dec 1999, Daniel Veillard wrote: > On Tue, Dec 21, 1999 at 11:48:49AM +0000, Dan Brickley wrote: > > > > See appendix for a first take at RDF/XLink mapping... > > I drafted it with some help from Ralph Swick but it's definitely > not a good mapping, I would be very grateful if people from the > RDF Interest list could suggest a better one. Ora Lassila suggested > using some reification around the main Source -> Target relations > to add all the links related metadata. That seems a good starting > point toward a nice mapping. It would be good if the result would not > seem awkward as a base to express RDFs constructs in terms of XLinks. > I have the sense (and it seems I'm not the only one :-) that a > bidirectionnal mapping is possible, I would definitely feel better if > this turned out to be true ! > > any taker ? I share your suspicion/hope that bidirectionality is feasible. The current mapping, as you say, doesn't feel quite right. Now that we have QName'd 'roles' for XLink link types, I'm wondering whether we could have a simple interpretation of typed links that (in the basic case) used the Qualified-named 'role' as the RDF property type connecting the 'from' and 'to' resources. Something like this raises prospect that we could mine a web of XLink documents for RDF assertions that could be slurped into an aggregated RDF database... Dan
Received on Tuesday, 21 December 1999 08:03:55 UTC