- From: Dan Connolly <connolly@w3.org>
- Date: Tue, 04 Sep 2001 17:33:10 -0500
- To: pat hayes <phayes@ai.uwf.edu>
- CC: w3c-rdfcore-wg@w3.org
pat hayes wrote: > > >pat hayes wrote: > > > > > > >Brian McBride wrote: > > > > > > > > > > pat hayes wrote: > > > > > [...] > > > > > > Wait a minute. The subject is a URI, not a Resource, right? The > > > > > > Resource is what the subject (a piece of syntax) denotes, not the > > > > > > subject itself. > > > > > > > >That's what you'd think, coming from a logic background, but > > > >as Brian points out, RDF says the subject of > > > > Mary hit the ball. > > > > > > > >is a female person, not a word starting with 'M'. > > > > > > > >This is mother of all use/mention bugs, IMO. > > > > > > > >cf > > > > > > Indeed. If we have to take this seriously then I withdraw the model > > > theory, since it can't possibly work. > > > >You mean the reification part of the model theory, right? > > > >The rest is fine. > > No, I meant it all. But I spoke hastily. I guess anything can be a > symbol, so maybe my mother *could* be in a grammar. Hang on... there are two parts to this issue: (1) sloppiness in the way the RDF core is specified, and (2) sloppiness in the way RDF reification is specified. Regarding (1), I suggest we regard all these use/mention bugs in the RDF M&S spec as just that: use/mention bugs; and let's clean them up. The spec is so rife with them that I favor pretty much starting over... with the model theory draft. In this model theory, we have a fairly ordinary syntax for RDF (whose components are ordinary syntactic symbols, though the "anonymous" terms are a bit novel). And in each interpretation of an RDF document/formula, we have a domain of discourse that might contain mothers, backhoes, etc. This is largely consistent with the way RDF has been deployed; this is the way people use it and the way the code works. It's just an editorial fix. Regarding (2), the damage is so extensive that I'm not sure it's recoverable. The sloppiness in the spec has propogated into running code to the extent that there's really not much value in changing the specification associated with the names http://www.w3.org/1999/02/22-rdf-syntax-ns#subject http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate http://www.w3.org/1999/02/22-rdf-syntax-ns#object because the code is already deployed, and we might as well make up new names if we're going to re-deploy the code. I'm quite happy with the way the draft model theory (pretty much all the drafts) deal with (1); with regard to (2), the drafts take a sane approach, but an approach that is inconsistent with practice. Again, I think this is why you were actioned to produce a MT draft without reification. [...] -- Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Tuesday, 4 September 2001 18:33:14 UTC