From: Stefan Decker (stefan@isi.edu)
Date: 12/03/02
Hi Jos, At 09:06 AM 12/3/2002, Jos De Roo wrote: >Hi Stefan, > >this is a bit of a reply (while having holiday) to your nice paper >http://www.daml.org/listarchive/joint-committee/att-1254/01-bubo.pdf thanks al ot! >... > > > Three systems try to implement OWL using a LP-based approach: > > The Euler Proof Mechanism [14] by Jos De Roo of Agfa and Tim > > Berners-Lees Closed World Machine (CWM) [2] correspond very > > closely and use the same syntactic format for rules. However both > > do not consider data integration. They try to axiomatize everything, > > whereas we try to rely on the features of logic itself, such > > as implication to operationalize the transitivity of subclassof. Their > > axiomatization is not proven to be correct or complete, e.g. they do > > not capture the substituivity of sameIndividualAs and capture only > > one direction of hasValue. There is no formaly proven characterization > > of the inference algorithms employed by Euler and CWM. > > So it is unclear what they are actually doing. > >really ;-) Oops - this statement in our paper is actually not quite correct (we will revise it) - what was meant is that there is no model theoretic semantics for the Euler proof procedure, which makes it is hard to see how Euler relates to to existing LP and deductive database technology. But it would be an interesting task to define one. Our paper has also a flaw - what is missing is a correspondence theorem, connecting the DL model theory and the LP model theory - delivering a soundness and completeness result for a DAML+OIL or OWL sublanguage. Best, Stefan
This archive was generated by hypermail 2.1.4 : 12/03/02 EST