Last modified on 25 January 2009, at 06:18
|I have been using the OWL 2 Manchester syntax in protégé 4 beta almost exclusively and find very few gaps in general. The area of exception that is still not dealt with in the current specification is the extension of datatypes to capture the ISO healthcare complex data types (presumably by external reference to an ISO or HL7 xsd). Because of the nesting of primitive XML datatypes to form more complex data types, one is left with modeling in OWL full. It would be nice to be able to reference external schema definitions and leave these outside the DL reasoning but still be able to point a dataProperty to them as Ranges.
Not having this is a huge challenge for implementations in the more complex healthcare fields that require standards for structured vocabulary.
Seems to be resolved in this followup:
We just completed the HL7 workgroup meeting last week. As part of the meeting, myself and an HL7/OMG colleage reviewed the issues associated with the HL7/ ISO healthcare datatypes, the OMG Ontology Definition Metamodel (ODM) and the OWL 2 specification. The simplest route was to modify the ISO datatypes by adding a closure axiom of sorts (a datatype flavor) that prevents continued recursion and allows an expression to eventually lead to primitive XML schema datatypes, all of which can now be expressed in OWL. We still need to meet with OMG and correct the ODM specification so that attributes can be either object properties or datatype properties rather than just datatype properties but I do not have any further concerns for W3C at this juncture.