- From: pat hayes <phayes@ihmc.us>
- Date: Tue, 10 Jun 2003 11:25:42 -0500
- To: Graham Klyne <gk@ninebynine.org>
- Cc: w3c-rdfcore-wg@w3.org
>At 16:42 09/06/03 -0500, pat hayes wrote: >>While fixing a silly mistake in the MT document, I noticed the >>following. We require that RDF graphs contain no redundancies in >>the sense that the same triple cannot occur more than once in the >>graph. However, they can contain redundancies in the sense that a >>triple with a bnode in it can be duplicated with a different bnode, >>even though the resulting triples would look the same in a graph >>diagram. The resulting graph has no extra information in it, but >>this quirk allows an RDF graph to have infinitely many >>consequences. For example, a single triple >> >>a p b . >> >>has infinitely many consequences; >> >>_:x p b . >>a p _:y . >>_:z1 p _:z2 . >>_:z3 p _:z4 . >>-:z5 p _:z6 . >>.... >> >>where all these bnodes are distinct; see attached jpeg. >> >>My question is, does the WG feel that it might be worth ruling this >>out as a syntactic possibility? If this kind of bnode-duplication >>were ruled out, then the set of graphs simply entailed by any RDF >>graph would be finite. That would generalize the >>no-duplicate-triples condition implicit in our definition of a >>graph as a set, to treat triples which 'look' the same when you >>erase the bnode labels as though they literally were the same. > >From an implementation perspective, I'd be concerned about ruling >out redundant bnode triples as syntactically disallowed. I think >this would make the implementation of merging two >syntactically-valid graphs very much more difficult (but not >impossible). Yes. and after sleeping on it, I think its best not to tinker with the syntax at this stage in any case. And I am pretty sure that I can handle the problem without doing anything so drastic. So, Never Mind. Pat -- --------------------------------------------------------------------- IHMC (850)434 8903 or (650)494 3973 home 40 South Alcaniz St. (850)202 4416 office Pensacola (850)202 4440 fax FL 32501 (850)291 0667 cell phayes@ihmc.us http://www.ihmc.us/users/phayes
Received on Tuesday, 10 June 2003 12:25:56 UTC