- From: Tanel Tammet <tammet@staff.ttu.ee>
- Date: Fri, 17 Dec 2004 23:37:22 +0200
- To: Sandro Hawke <sandro@w3.org>
- Cc: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, www-rdf-logic@w3.org
Received on Friday, 17 December 2004 21:37:04 UTC
Sandro Hawke wrote: >>See above. >> >>The ability to derive consequences is the heart of a logic. If you can't >>get FOL entailment from your machinery you haven't done FOL. >> >> >.... > > >The goal, again, is for someone to be able to express the definition >of fam:uncle in RDF in a way which supports the obvious inferences and >causes no serious problems. If it's not clear yet how this design >does that, what would help clarify it? > > > I do not have much to add to this discussion. Rather, I'd propose another angle to look at the possibility to encode FOL in RDF. As far as I understand, RDF does not have universally quantified variables and is obviously decidable. RDFS has universally quantified variables in some specific formulas (taxonomy implications), but is still decidable. FOL, on the other hand, is undecidable. Hence you cannot encode FOL in RDF(S), using RDF(S) semantics and nothing more. On the other hand, you can write FOL sentences as bitstrings, hence you can obviously write them as RDF sentences too. However, such RDF usage would have no logical meaning (i.e. you need an external mechanism for derivations) and would not be especially interesting, Since this is pretty obvious, I was a bit puzzled by the issue you were discussing. Could you clarify a bit: what was the main idea of the question Sandro posed? Regards, Tanel Tammet
Received on Friday, 17 December 2004 21:37:04 UTC