Eric,
Thanks for this very informative response. I had not really
appreciated until just now that GRDDL was intended to be applicable
to arbitrary XML instance data as well, and this opens tremendous
opportunities (but also suggests a need to really beat the drum among
document standards groups !!!)
I wonder whether this general item (RDF embedding technologies)
deserves an explicit place in one of the workgroups. I don't think
the best home is Ontologies workgroup. Maybe the Process/Choreography
workgroup??
John
On Feb 10, 2006, at 9:15 AM, Eric Miller wrote:
On Feb 10, 2006, at 3:41 AM, John Madden wrote:
Daniel,
I spoke with Mark Musen about this project at the F2F, and we
(SNOMED) would be eager to work collaboratively with NCBO on this
as a demo. I'm coming out to the Protege short course next month
(funded by SNOMED) and I'll have some material for you by then.
(a) I think is to a large extent a "just-do-it" question, and we
definitely want to "do it". The interesting part to me is the
follow-up project of showing how OWL fragments built separately in
this way can be related to a more global ontology that may not be
quite so rigorously formal; that's (b).
For me, in many ways the hardest part is (c), a standard way of
encapsulating RDF-family markup in xml-schema dependent document
formats. (I know this takes us afield from "ontologies" per se.
Personally, I'd like to see GRDDL applied to these formats, but,
we've got to be realistic that clinical medical documents today
are almost never natively html, and GRDDL is/was rather html-
focused. Also, you might argue that these formats do not assume
web connectivity so it is insufficient to merely point to an
extraction transform somewhere out on the web, when that might not
be accessible; I'm not sure whether that really matters.)
To be clear, GRRDL is not specific to XHTML and can be used for
translating XML instance data as well. One can use GRDDL for
"schema annotation" and transform all of the instance data that
conforms to a particular schema into RDF. I think the following
provides some of the relevant bits that explain this
- http://www.w3.org/TeamSubmission/grddl/#ns-bind
The current GRDDL documents don't elaborate on this 'schema
annotation' use case in as much detail as I think it deserves but I
expect this issue will be addressed in future work. An exploration
of this 'schema annotation' approach using GRDDL in a different
context is here
- http://www.w3.org/2003/g/cc/demo
hmm... not sure this is enough to go on in and of itself :( but it
may be useful.
I've spoken with the OpenDocument group about an RDF embedding
standard and they're thinking about it but haven't come to any
firm proposal. I also asked Norm Walsh about it a few months ago
and he had it on his issue list for DocBook, but again I'm not yet
aware of a definitive solution. For HL7-CDA, things are hazy right
now.
Daniel, I'd be interested in talking over details with you when
I'm out in Stanford.
The real challenge is modeling the semantics of HL7 and creating
the appropriate transformations. Please report back any discussions
to this list as I (along with many others it seems :) am interested
in this area.
--
eric miller http://www.w3.org/people/em/
semantic web activity lead http://www.w3.org/2001/sw/
w3c world wide web consortium http://www.w3.org/
