- From: Sergey Melnik <melnik@db.stanford.edu>
- Date: Fri, 15 Jun 2001 18:36:48 -0700
- To: Aaron Swartz <aswartz@upclink.com>
- CC: Brian McBride <bwm@hplb.hpl.hp.com>, rdf core <w3c-rdfcore-wg@w3.org>
Aaron Swartz wrote: > > On Friday, June 15, 2001, at 06:49 PM, Sergey Melnik wrote: > > > IMO reification has a chance to be introduced as a pure model feature > > not intrinsically bound to any vocabulary > > Why do you want this? Is there a reason every RDF application > has to understand reification? I hope not, because that will > make RDF very confusing. I did not mean to declare reification a mandatory feature. Rather, I'd define it as a separate "layer", so that whoever needed reification could rely upon a recommended practice. > > In this perspective, it would not go under 'vocabularies'. Similarly, > > namespaces and literals are another two model features that might go > > into one of the layers in the 'core'. > > To my knowledge, you still have not explained how namespaces fit > into the model. If I missed this somewhere, please send me a > pointer. I'm quite curious what you mean. In fact, some time ago I mentioned "Slim RDF" on the RDF Interest list http://lists.w3.org/Archives/Public/www-rdf-interest/2001Feb/0090.html). There, namespaces were introduced as part of the model. The reason behind it that currently, given a resource, applications cannot determine the namespace of the resource (they can only guess by looking for "#" or last "/"). Several people who used the "RDF API" for implementing editors for RDF Schemas (including Michael Sintek who collaborated on Protege) complained that the API did not offer a way of reliably determining the namespace of a resource. Since a year or so, the API supports that feature, effectively assuming that namespaces are part of the model. Sergey
Received on Friday, 15 June 2001 21:23:55 UTC