some thoughts about the OWL WG comments

Alistair, Sean, all,

I promised (in our post-telecon discussion) to provide some thoughts on 
the issues raised by Peter on behalf of the OWL WG [1].

Guus


NOTATIONS

[[
   The OWL WG notes that one portion of SKOS (Notations) uses custom
datatypes.  Although these seem to be benign, because RDF and OWL allow
extra datatypes, the use of these datatypes is not likely to be
supported by many tools.  The presence of extra datatypes may cause
difficulties in some tools, which may just reject SKOS documents that
have these datatypes.
]]

Sean already pointed out that we can indicate in the document that 
user-defined datatypes are only explicitly needed when one needs 
multiple notations (so only in Sec. ). In addition I would like to add 
the tools should not reject SKOS documents containing user-defined 
datatypes, see the OWL reference section on datatype reasoning [2], in 
particular the last sentence:

[[
   Unrecognized datatypes should be treated in the same way as 
unsupported datatypes.
]]


LEXICAL LABELS

[[
     objects as values of data property (example)
     suggestion: don't do this
]]

Note sure what this comment refers to. All examples in 5.4 [3] use 
literals as values of label properties. The skos-xl extension handle 
object as values of label properties in an OWL-DL consistent manner.


DOCUMENTATION PROPERTIES

[[
     using literal in object property (examples)
     suggestion: don't do this
]]

We define the skos:note and its subproperties currently as 
owl:ObjectProperty. From an OWL Full perspective this is fine 
(owl:DatatypeProperty is a subproperty of owl:ObjectProperty in OWL Full 
[4]), but for OWL DL this is a problem. Part of the problem is that OWL 
forces you to make a choice between either object or datatype property, 
and we do not want to force this choice upon SKOS users. My proposal 
would be to follow a "least-commitment" strategy and change skos:note to 
be just an rdf:Property. This does not make it OWL-DL compliant yet, but 
allows people who want to use it within OWL DL to add a triple with the 
required OWL property type. So instead of being OWL-DL inconsistent it 
becomes OWL-DL incomplete.


[[
   use of rdf:value (example)
   suggestion: don't use rdf:value
]]

This refers to example 25 [5]. I note that rdf:value has no particular 
semantics and is mainly a usage convention (and in practice is actually 
not used a lot). I suggest to change the example to use a user-defined 
property to refer to the value. I also suggest (but this is independent 
of the OWL WG comment), to add another property statement to illustrate 
why this pattern is used at all.

Whether this also implies changes to the Primer I'm not sure. We may 
have the pattern with the blank node to have two variations: one with a 
custom value property and one with rdf:value.

I also note that we use rdf:value a lot in the namespace file [6] to 
specify change notes. My proposal is to drop these change notes altogether.


[1] http://lists.w3.org/Archives/Public/public-swd-wg/2008Oct/0059.html
[2] http://www.w3.org/TR/owl-ref/#DatatypeSupport
[3] http://www.w3.org/TR/2008/WD-skos-reference-20080829/#L1409
[4] http://www.w3.org/TR/owl-ref/#OWLFull
[5] http://www.w3.org/TR/2008/WD-skos-reference-20080829/#L1812
[6] http://www.w3.org/TR/2008/WD-skos-reference-20080829/skos.rdf