- From: Gerd Wagner <wagnerg@tu-cottbus.de>
- Date: Thu, 18 May 2006 15:19:15 +0200
- To: "'Peter F. Patel-Schneider'" <pfps@inf.unibz.it>
- Cc: <public-rif-wg@w3.org>
> > The proposal may not be sufficiently explicit about this, > > but it states that modeltheoretic satisfaction gives > > the meaning to conditions. > > Yes, but where is this tied to the mappings? This has not been made explicit in the proposal. So, let's do it. [Harold and Michael, we are waiting for your contribution to this.] It seems clear to me, that if the RIF core has a model-theoretic semantics in the form of a preferred/intended model operator Mod assigning a set of intended models to a set of formulas (as a borderline case, giving you what you prefer, all models may be considered intended models), then we want to have that for any RIF condition formula C Mod(C) = Mod'(M[C]) where M is the mapping from RIF to the target language, and Mod' is the adapted model operator definition working on formulas of the target language. Considering also the inverse mapping N from the target language to RIF, we get the additional requirement that Mod(C) = Mod(N[M[C]]) which is, however not sufficient to guarantee semantic equivalence. > > The proposal mentions the option of typing terms and > predicates/atoms. > > So, I just made use of this option (in the spirit of the proposal). > > Of course, you are free to suggest another style of typing... > > How about adding K and A modal operators? This would be another extension that we could think about, if we find arguments to justify it. > > >> The extended proposal syntax with optional typing also > > >> allows a faithful inverse mapping of typed atoms to SWRL. > > > > > > Oh? Which atoms? All of them? > > > > Yes all of them (modulo some subtleties concerning restrictions > > and the generic datatype rdfs:Literal). > > Even predicate applications with 5 arguments? No, you are right, in some cases, like this one, we may need to apply special transformation techniques (which are well-known for this reduction from n-ary to binary), or we may even have to give up. As I said already, typically, mappings will be partial. -Gerd
Received on Thursday, 18 May 2006 13:19:28 UTC