- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Sat, 20 Dec 2008 15:32:50 +0000
- To: Jos de Bruijn <debruijn@inf.unibz.it>
- CC: RIF WG <public-rif-wg@w3.org>
Jos de Bruijn wrote: > I was thinking a bit more about the problem of the negative guards and > the static rule set we wish to provide for OWL RL. > We established that without negative guards, we cannot provide a finite > static ruleset, because the rules would depend on the literals actually > being used in the ontology. > > Now, a property of the OWL 2 language is that there is a strict > separation between the object and data domains, both in the syntax and > the semantics. For example, it is not possible to assert that an IRI is > a member of a datatype, nor is it possible to assert equality between > and IRI and a concrete data value. > Most importantly, the values of object properties can only be IRIs and > the values of data properties can only be literals. And, in OWL 2 RL we > only need to take individuals into account that are explicitly > represented using IRI or literals. > > Therefore, a statement like t[rdf:type -> DT], where DT is a datatype > can never be derived if t is not a literal. Consequently, we only need > to do type checking of the form at the bottom of section 4.4.2 of [1] > for literal values. > > So, I believe that a restricted form kind of negative guard, namely one > that is restricted to the domain of literals (e.g., > isNonIntegerLiteral), is sufficient for this static ruleset. Agreed. [I think when we discussed isNotType we mentioned the possibility of limiting it to isNotTypeLiteral.] Dave
Received on Saturday, 20 December 2008 15:33:39 UTC