Comments on the RDF Mapping

I have a number of comments to make on the WSDL 2.0: RDF Mapping
http://www.w3.org/TR/wsdl20-rdf/

Actually, the mapping is welcomingly straightforward, with the number of
design 'tricks' kept to a minimum. For example, representing properties
with classes appears to be a reasonable compromise between a literal
(pedantic) mapping and something (over-clever) utilizing the full
descriptive capabilities of OWL. Similarly, the number of 'invented'
URIs has been kept to a minimum. We appreciate the 'Keep it Semantically
Simple' approach.

The Customer base for the RDF mapping is likely to be the semantic
annotation for WSDL working group, who may use the mapping as a
framework on which to hang additional semantic annotations.

Moving the maping to another working group doesn't seem to be an ideal
option, it should be for the web-services description group to define
the semantics of wsdl itself. An annotations group should be concerned
only with extensions.

I like the (intermediate) solution adopted to reference XML schema
components (problematic because although elements and types (for
example) are defined in the same schema target namespace they
nevertheless exist in disjoint namespaces.

I wonder if additional disjoint relationships be defined in the OWL. ie.
#HTTPBindingWithCookies disjoint with the WSDL 2 SOAP binding?

Finally, it would be good to see the mapping to RDF realized as an XSLT
transform.

This is a nice piece of work, and I believe the community at large would
benefit from keeping the WSDL 2.0 RDF mapping within the recommendation.

Steve Battle (Hewlett Packard)

Received on Tuesday, 31 January 2006 11:59:34 UTC