- From: Michael Kifer <kifer@cs.sunysb.edu>
- Date: Thu, 04 Jan 2007 10:43:54 -0500
- To: Dave Reynolds <der@hplb.hpl.hp.com>
- Cc: Chris Welty <cawelty@gmail.com>, RIF <public-rif-wg@w3.org>
Dave Reynolds <der@hplb.hpl.hp.com> wrote: > > What I don't understand is what people want out of this slotted notation > in the first place to know whether the open world assumption would > negate its value to them. > > It's certainly the case that one of our major support costs is users > approaching RDFS/OWL as if it were object oriented and implicitly making > closed world assumptions and getting burned. So I'm keen that whatever > choices are made here are at least spelled out extremely clearly. We have two options (at least): 1. Not to introduce signatures into the core. 2. Introduce signatures into the core and require that p(foo->a), p(foo=>b) implies a:b. This doesn't preclude CWA in the dialects because for CWA one needs to introduce the notion of well-typed models. The notion of a well-typed model is an elaboration on the notion of intended models where the above property is satisfied. For instance, there would be no well-typed models for {p(foo->a), p(foo=>b)}. This is how it is defined, for instance, in F-logic under the LP semantics. --michael
Received on Thursday, 4 January 2007 15:44:02 UTC