- From: Jos De_Roo <jos.deroo.jd@belgium.agfa.com>
- Date: Fri, 15 Mar 2002 00:46:27 +0100
- To: pfps@research.bell-labs.com
- Cc: "WebOnt WG <www-webont-wg" <www-webont-wg@w3.org>
> All triples are assertions. Oh no!!! let's remember what Pat said a while ago [[[ This was in response to Jim's remark that we need a way for one ontology to refer to a part of an ontology and commit to it. My point was that we should distinguish between the reference to a part of an ontology, and the commitment to the part so referenced, so that one can refer to a part of an ontology without automatically thereby assenting to it. One reason for emphasizing this point is that there are several use cases already (notably Jos deRoos' implementation of N3) where the 'grouping' technique is used to gather together a set of antecedents of an implication, so that one can say [all this stuff] log:implies [this other stuff]. More generally, however, I would suggest that we take care to keep functionally distinct aspects of the language as distinct as possible, and that referring to/pointing to/whatever some ontology ought to one thing, and any speech act (assenting, asserting, denying, questioning, expressing doubt about, saying it is connected to foo, saying it entails foo....) involving it should be something else. It would be OK to have a default case where if you just 'say' it without any further comment then that is taken to be an assertion (assention?), but it ought to be *very* easy to override that assumption. And I think it would be best to have an explicit 'we include this here' marker, like DAML's 'import'. ]]] let's indeed keep things as distinct as possible and we have URI's to do so, unambiguously, decoupled -- Jos
Received on Thursday, 14 March 2002 18:47:05 UTC