- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Wed, 28 Nov 2007 11:06:17 +0000
- To: "Web Ontology Language ((OWL)) Working Group WG" <public-owl-wg@w3.org>
With regards to ISSUE-72 ISSUE-63 I thought the following tests may be helpful. Positive entailment test: ========================= first:A rdf:type owl:Class . first:A first:ap "an annotation" . first:ap rdf:type owl:AnnotationProperty . entails first:A rdf:type owl:Class . _:a rdf:type owl:Thing . first:A first:ap "an annotation" . first:ap rdf:type owl:AnnotationProperty . i.e. an ontology including an annotation entails itself, feel free to reformulate in whatever notation. And Negative entailment test: ========================= first:A rdf:type owl:Class . first:A first:ap "an annotation" . first:ap rdf:type owl:AnnotationProperty . does not entail first:A rdf:type owl:Class . _:a rdf:type owl:Thing . first:A first:ap "a different annotation" . first:ap rdf:type owl:AnnotationProperty . i.e. the mechanisms for dealing with entailments do not permit new annotations to be introduced. Both of these conform to OWL 1.0. Ian states that OWL DL implementations do not get this right. If that is so, we should decide whether to: a) reiterate the previous decisions and correct the test suite by adding such tests, to point out to OWL DL implementors what they are expected to do, and to help them build conformant systems or b) change the specifications to conform to the implementations - noting that OWL Full reasoners, such as Jena's rule engine, do correctly implement the current specification. Hence we have an obligation to OWL Full implementators not to change the spec in a way that makes previously correct implementations incorrect. Without deciding on our stance with respect to OWL 1.0 annotations, it is pointless to try and discuss axiom annotations. Jeremy
Received on Wednesday, 28 November 2007 11:06:47 UTC