- From: Jeff Heflin <heflin@cse.lehigh.edu>
- Date: Wed, 18 Sep 2002 17:43:11 -0400
- To: pat hayes <phayes@ai.uwf.edu>
- CC: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>, www-webont-wg@w3.org
pat hayes wrote: > > I agree with Peter, but how about the following suggestion for > re-phrasing this in terms of entailment. If an ontology A contains > [import B] (in whatever notation turns out to be appropriate) and if > B + A entails C then A entails C, where '+' means whatever is the > appropriate 'conjunction' (merging) operation on ontologies. (For > bare RDF that would be a triples-graph merge, but OWL might have a > more sophisticated notion involving the addition of some extra stuff > to keep the merge tidy; whatever. The point here is that we will need > this notion to be defined for ANY web ontology language, so whatever > it is, 'imports' refers to that.) That's sounds fine to me, but I don't think its any different from my original proposal which could be restated as "if ontology A contains [import B] and B entails P then A entails P." Doesn't all the entailments of B+A fall out that way? It's the end of the day and I'm tired :-( However, if you agree with Peter and I agree with this, maybe we've all been in violent agreement all along? > Now, one can take this to be a constraint on what 'entails' means, or > one can take it instead to be a specification constraint on behaviors > of reasoners: that in order to be complete with respect to > entailment, a reasoner needs to somehow 'merge' the content of the > imported ontology (and hence, the transitive closure of other > imported ontologies that they might require in order to be complete). > But there are many ways one might do that: by copying and caching B > into A, by retrieval from B during inference from A when needed, if > that is possible, or even by querying B using a query language (under > some circumstances). But that decision - or indeed, the decision to > completely ignore B - can be left to the implementation. An engine > that makes no pretensions to completeness - and I hereby predict that > almost all deployed reasoners will not make any such claims, since > they will be largely meaningless in a real web context - can choose > to ignore imports statements, or to treat them very casually, for > example, and still conform to this kind of spec. That's fine. > I think any spec that *requires* imports to be implemented by a > syntactic copy is simply not acceptable and will not be used; and > that in any case it is not necessary. And all such interpretations > run foul of the syntactic vagaries of XML and RDF. I agree whole-heartedly. > > Pat > > -- > --------------------------------------------------------------------- > IHMC (850)434 8903 home > 40 South Alcaniz St. (850)202 4416 office > Pensacola, FL 32501 (850)202 4440 fax > phayes@ai.uwf.edu > http://www.coginst.uwf.edu/~phayes
Received on Wednesday, 18 September 2002 17:43:16 UTC