- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Tue, 16 Sep 2003 11:58:21 -0400 (EDT)
- To: jjc@hpl.hp.com
- Cc: www-webont-wg@w3.org
From: Jeremy Carroll <jjc@hpl.hp.com> Subject: S&AS question datatype URI syntax Date: Mon, 15 Sep 2003 15:11:58 +0300 > This is a comment on: > http://www.w3.org/TR/2003/CR-owl-semantics-20030818/ > sections 2 and 4. > > > I had assumed that the following is not legal OWL DL > > eg:prop rdf:type owl:DatatypeProperty . > _:a rdf:type owl:Thing . > _:a eg:prop "foo"^^<http://www.example.org/some/uri>. > > because the URI used in the typed literal is not a datatypeIDs (either > xsd:XXX rdfs:XMLLiteral or maybe a user datatype declared as such - I have > lost the plot on the resolution of the last possibility). > > However I can't see how S&AS rules out the above graph. > > Have I misread this, or is it intended, or is it a bug? > > Jeremy Bug or feature is in the eye of the beholder. The current situation is that an OWL dataliteral in the abstract syntax is either an untyped literal or a typed literal. The type of a typed literal is a URIReference (Note the typo, just fixed in my copy.) not a datatypeID. Under this reading, there is no need that there be a datatype typing triple for them. Is this a bug? I don't know. Changing it would not be hard, but, again, should require a working group decision. The change would be along the lines: 1/ Require that the types of typed literals in the OWL abstract syntax be datatypeIDs. 2/ Change the mapping for literals in 4.1 to have a case for typed literals that would be of the form. Syntax lexicalForm^^datatypeID Transformation datatypeID rdf:type rdfs:Datatype . datatypeID rdf:type rdfs:Class . [opt] Main node lexicalForm^^datatypeID Productions for untyped literals would also be needed. peter
Received on Tuesday, 16 September 2003 11:59:51 UTC