Re: RDF binding, Java::IMS MD-Interoperability

(Copying the RDF and DC-Architecture lists; see http://www.w3.org/RDF/
and http://www.dublincore.org/groups/architecture/ for more
context; see also
http://lists.w3.org/Archives/Public/www-rdf-interest/2001Mar/0248.html
for my earlier heads-up message on this topic)


On Thu, 29 Mar 2001 Forum_Agent%CLN5-3%COLLEGIS@Collegis.org wrote:

> Posted by:  Mikael Nilsson
> 
> On Thu, 29 Mar 2001 Forum_Agent%CLN5-3%COLLEGIS@Collegis.org wrote:
> 
> > IMS-land, but will do whatever I can to help with this effort. Is it OK
> to
> > forward details to other mailing lists, projects, where I can probably
> > find RDF implementors who would be keen to help out?
> 
> Please do. We need all the critical eyes and helping hands we can get.

OK, most of the RDF/DC people who should see this are copied. There may
be some modelling and schema design issues of interest to the DC
Education group too (http://www.dublincore.org/groups/education/), but
for now I think we need to focus on low level engineering / technical
concerns. I'm copying Jon Mason and Stuart Sutton of DCMI Education WG
for info. BTW I have no knowledge of how this relates to the
IEEE/LTSC/LOM <-> DCMI announcement of http://dublincore.org/news/pr-20001206.shtml
-- to my mind this current thread suggests some technical options that
will help us turn this press release into a practical proposal.

> 
> > I should also mention that the Dublin Core Architecture WG is in the
> > process of cleaning up DC's RDF representation, so we would be delighted
> > to use this as a practical proposal when discussing interoperability
> > between Dublin Core and IMS. If someone gives me the go-ahead on this
> > I'd like to give the dc-architecture list details of the proposal ASAP.

> 
> This would of course be wonderful! The perfect dream would be if we could
> get DC and IMS compatible on a syntactical level in RDF. This should not be
> all that impossible, if we could reuse some basic datastructures early in
> the design process, i.e., now. Please, if you can, give some details on
> what DC plans to do in their cleanup.


OK, here's a brief summary of where we're up to with Dublin Core in RDF. 

Simple DC: the 15 DC elements are modelled as RDF properties in the
namespace 'http://purl.org/dc/elements/1.1/'. RDF applications are free
to use this in various ways (eg. the RSS newsfeed spec, see
http://purl.org/rss/, PRISM etc define RDF-based formats that
make use of these properties). At this level, DC-in-RDF imposes no
constraints regarding XML format; DC properties can be mixed in with an
syntactic representation of RDF.

In addition, we have a DC spec that defines a simple utility XML format
for DC records that fit a certain constrained profile. Dave Beckett has done
most of the work on this, see:
http://www.dublincore.org/documents/2000/11/dcmes-xml/ 

Most recently, Stefan Kokkelink and Roland Schwanzl have produced a new
working draft for the DC Architecture WG that improves upon the old
proposal (http://www.ukoln.ac.uk/metadata/resources/dc/datamodel/WD-dc-rdf/) 
in a number of ways. (*apologetic aside*: I'm aware that the DC web
pages describing this need updating, as does the old WD-dc-rdf page; this is
in progress). The new draft makes better use of RDF's built-in
facilities (the sub-property relation, and the type system). Our earlier
approach involved instance data carrying meta-information that is better
provided at the schema level.

The DC RDF draft is available at:

http://www.mathematik.uni-osnabrueck.de/projects/dcqual/qual21.3.1/

Please copy mailto:dc-architecture@jiscmail.ac.uk on any comments / feedback
 

> > Let me know how best I can help,
> 
> If you are fluent in RDF (and I guess you are), give us design input. Help
> us formulate problems that need to be addressed for interoperability with
> for example DC, or classical design problems when interoperating with
> information models such as LOM or DC and XML DTDs. Or simply tell people
> about the effort. All help appreciated!

I hope I've at least managed the latter. Regarding XML DTDs and
Schemas, I believe the message is simple: there will be _lots_ of DTDs
and XML Schemas, each reflecting local concerns. Our common RDF model is
an attempt to bind together data from various such environments.

As for detailed design input, I need to find out more about the current
nature of the IMS work before wading in. A broad-brush perspective
on the problem can be found in the DC-Architecture presentation at 

http://dublincore.org/workshops/dc8/agenda-resources.shtml
-> http://dublincore.org/workshops/dc8/DCMIArchitecture2/

specifically the observation:

	"Finding documents requires describing documents...
	 Finding documents requires describing people...
	 Finding education-related documents requires describing additional 
	 characteristics of those people and documents",

..ie. I'm trying to say that our problem is that all metadata
applications overlap, and that it is the job of the DC Architecture WG
to explore mechanisms by which DC can work well with other metadata
specs with a minimum of coordination. It also suggests that our
discussions should be grounded in discovery, ie. we talk about the kinds
of questions we might want such mixed metadata to allow us to answer.

Also -- the DC/RDF draft mentions this, but I'll mention here explictly
-- you might want to take a look at the "VCard in RDF" Note recently
published by the World Wide Web Consortium. 
http://www.w3.org/TR/#Notes -> http://www.w3.org/TR/vcard-rdf as this
discusses an RDF vocabulary for representing certain characteristics of
people and organisations.


> 
> About the posted documents: they will not parse... please be patient with
> this, and try to grasp the general ideas involved. Next week things will
> perhaps start parsing...
> 
> /Mikael

OK, I look forward to it :)

BTW the "RDF Sandbox" at http://www.w3.org/RDF/Implementations/SiRPAC/
is a useful tool for implementation checking, especially now it also
includes a graphical visualisation tool.

cheers,

Dan

-- 

DC-Architecture co-chair


> ## To send your response to the forum, change the mailto address to:
> ## IMS_MD-Interoperability@collegis.org
> ## The current message and attachments may be found at
> ## http://ims.collegis.org/md-interop.nsf
> ## This document can be found at the following URL:
> ##
> http://ims.collegis.org/md-interop.nsf/docid/725EFF03DE45790685256A1E00627453
> ## Please do not delete the following two lines.
> ## ParentDocID:725EFF03DE45790685256A1E00627453
> ## FormType:MainTopic
> 
> 
>

Received on Friday, 30 March 2001 08:48:10 UTC