Re: schemaMapping issues breakdown (issue 6)

Jacek,

Your suggestion for mappings (I'd say "no, yes, both".)
pretty much matches our original intent.  Since the data
mapping was based on the two schemas (e.g., in XSD and OWL)
we called it schema mapping, even though it is meant to be
applied to data instances at run time.

One of my prior students used the Lumina discovery tool to
find "real" semantically matching services.  Without mapping,
only a tiny percentage could be invoked.  With fairly simple
mappings all of them could.  The mappings provide a way
to "connect to the semantic bus" so to speak.

Although, the mapping aspect of SAWSDL may not be as elegant
as the rest, I think it is a useful and practical feature
to include.

Thanks,
John

P.S.  I also see Rama's point regarding for example, XSLT
mappings.  Since XSLT is Turing complete, we can't do much
with it other than invoke it (e.g., don't know how useful
it would be in discovery).

Received on Monday, 29 May 2006 22:34:19 UTC