proposal for OWL datatypes

[This message is a reply to a previous message that I sent out that is
closely related.]

0/ Today the working group voted to broaden OWL datatypes to allow for
   ``unsupported'' datatypes (Ian's wording) or ``unrecognized'' datatypes
   (RDF wording).  This essentially means letting OWL use datatypes for
   which an OWL reasoner (``unsupported'') or the OWL formalism itself
   (``unrecognized'') does not understand the value space or the lexical to
   value mapping.  These datatypes would be interpreted by an OWL reasoner
   in the same manner as unrecognized datatypes, i.e., lexically identical
   literals would be equal and lexically different literals would not be
   known to be either equal or unequal.

   This is essentially Part G of the OWL DL Syntax that was approved today.
   I have expanded the explanation of the proposal slightly and allowed
   unrecognized datatypes into OWL Full as well.

   One effect of this is that one could say (in any species of OWL)
   something like

	ex:John rdf:type owl:Thing .
	ex:age rdf:type owl:DatatypeProperty .
	ex:rational rdf:type rdfs:Datatype .
	ex:John ex:age "7/9"^^ex:rational .

   without it being a syntax error (as it currently would be).

   OWL reasoners could claim to handle this datatype, or not, as the case
   may be.

1/ I propose that an OWL reasoner may choose not to support some of the OWL
   built-in datatypes or even rdf:XMLLiteral, but that it should be easy to
   determine which datatypes any OWL reasoner supports.

2/ I propose that all OWL reasoners be required to support the following
   TWO datatypes:

	xsd:integer
	xsd:string


Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Thursday, 13 March 2003 14:13:39 UTC