- From: Ian Horrocks <horrocks@cs.man.ac.uk>
- Date: Thu, 4 Mar 2004 14:01:39 +0000
- To: "Yuzhong Qu" <yzqu@seu.edu.cn>
- Cc: "Eric Jain" <Eric.Jain@isb-sib.ch>, <www-rdf-logic@w3.org>
On March 3, Yuzhong Qu writes: > > > > Your example falls into the second case. It seems that there is no > > > way by using OWL DL. > > > > Thanks. Would be interesting to know whether this was left out by > > oversight or intentionally, it seems like a rather basic (but perhaps > > problematic?) concept to me... > > > > Based on my knowledge, some Description Logics include the constructor for role complement, while some others don't. > > OWL DL doesn't support role (i.e. property) complement, the language designers (of OWL DL) may consider the issue of computational complex (as well as the decidability problem). > > Maybe, somebody can give us the right answer. The decidability and complexity experts (Uli Sattler and Carsten Lutz) tell me that adding role negation to OWL DL would not make it undecidable (because we would still be inside the C2 fragment of FOL), but that role negation is known to be a source of complexity (reasoning becomes NexpTime-hard, even for much simpler logics than OWL DL, if we assume that we have an unbounded number of different role names available). Ian > > > Yuzhong Qu >
Received on Thursday, 4 March 2004 08:59:37 UTC