- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Mon, 22 Sep 2003 13:30:15 +0300
- To: www-webont-wg@w3.org
I have put more tests in, completing the actions related to adding tests from Thursday telecon. I can't get to the test cases doc, but I have checked it in - I always get an old copy. I have in addition - - tried to add these tests to cover some known deficincies in coverage of OWL Test Cases, see particularly the new tests on TransitiveProperty, SymmetricProperty, backwardCompatibleWith. - also added a test for recursive imports. I will start separate threads on: - imports-012 - response to brickley - Ontology and semantic layering From the change log (note G.5 should have been G.4.1 etc - will fix soon). [[ G.4. New Tests Illustrating CR Issue Resolutions (Proposed or Approved) The tests mentioned here are associated with technical changes or nonchanges, in [OWL Semantics and Abstract Syntax], as agreed by the WG, in response to comments during the CR. G.5. owl:Ontology Changes Many of the tests in the Candidate Recommendation of 18th August were in fact incorrect, when compared with the mapping rules in the corresponding Semantics and Abstract Syntax. The first two mapping rules together made it compulsory to include a triple with predicate rdf:type and object owl:Ontology. This was unintended, and has been fixed. New tests Ontology-001, Ontology-002, Ontology-003 and Ontology-004 explore this, with the premises of tests Ontology-001, and Ontology-004 and the conclusions of tests Ontology-001, Ontology-002 and Ontology-003 having been modified to conform with the old rules. With no change to the mapping rules, many tests would have needed such modification. The new mapping rules still require such a triple when the ontology uses an ontology property or annotation property. This is exhibited in the tests backwardCompatibleWith-001 and backwardCompatibleWith-002. Further, since there was implementor comment concerning this issue with respect to owl:imports, we added imports-011 following suggestion of Jim Hendler, which includes the pertinent rdf:type owl:Ontology triple from imports-001. G.5.1. User defined Datatype URIs There is on-going discussion as to whether URIs used for user defined datatypes need to have an explicit type triple, tests I5.8-013, I5.8-014, I5.8-015 and I5.8-016 reflect this issue. Peter and Jeremy agree on the first three and agree that the last is not as stated in S&AS. Jeremy is arguing that S&AS should be changed and I5.8-016 should be correct. G.5.2. Superfluous owl:ObjectProperty triples The August 18th CR of OWL Semantics & Abstract Syntax, includes the following two mapping rules: individualvaluedPropertyID => individualvaluedPropertyID rdf:type owl:ObjectProperty . individualvaluedPropertyID rdf:type rdf:Property . [opt] and ObjectProperty(ID ... => ID rdf:type owl:ObjectProperty . [opt if one of the last three triples is included] ... Unfortunately, for ontologies that actually use a declared object property, which was also an owl:InverseFunctionalProperty, owl:SymmetricProperty or owl:TransitiveProperty, the first rule took precedence over the intent of the second. This has been changed. The following new tests have been added including syntax which falls foul of a strict reading of the August 18th rules, but now are in OWL Lite or OWL DL: SymmeticProperty-002, SymmeticProperty-003 and TransitiveProperty-002. They have also been chosen to exercise semantic features that were not adequately covered by the August 18th version of this document. ]] [[ A test for recursive processing in owl:imports was added: imports-012 G.6.1. Concerning owl:Class and rdfs:Class There were continued comments concerning the problems of owl:Class and rdfs:Class. The following tests were added in the hope of clarifying the situation: Class-001, Class-002, Class-003 and Class-004, ]]
Received on Monday, 22 September 2003 07:31:10 UTC