Hello all. Raymond raises a significant issue about handling XML-based
and RDF-based metadata. I'm part of a group that has been working to
develop an ISO standard for PDF/A, which aims to specify use of PDF in
a way that supports long-term management and preservation. The current
draft calls for PDF/A to use Adobe's XMP spec for metadata. XMP is
based on RDF and offers many advantages, including embedding as plain
text within PDF files (it also has some current disadvantages, including
limits on machine validation against schemas). If XMP remains in PDF/A
draft, if the standard is approved, and if the standard achieves some
broad use (all appear likely at this point), the use of RDF for metadata
For more information, about PDF/A see
http://www.aiim.org/standards.asp?ID=25013 . Adobe has also just
released a new version of the XMP spec--see
Library of Congress
>>> [log in to unmask] 2/17/2004 8:49:34 PM >>>
Since MacKenzie brought up the IU and our work on METS to IMS and RSS
and the relationship to RSS, let me jump in a bit with more later. I
also don't claim to give any definitive answers -- but I was doing a
lot of thinking about RDF a while back and wrote up my thoughts about
RDF coming from a non-RDF XML perspective. I believe that it's still
accurate reflection of my current thoughts -- and the writeup might be
helpful to you. It's on my wiki:
I'd be very interested in any reactions you might have.
One of the big issues for me is to figure out how to handle both
XML-based metadata and the (presumably) increasing amount of RDF-based
metadata we'll begin seeing. Let me get back to you all when I have
more thoughts on the topic.
MacKenzie Smith wrote:
> Hi Jinfang,
>> anybody here can talk about something between METS and RDF? how are
>> different? since they both can combine different kinds of metadata
>> Besides the difference that METS can encode the structure of
>> resources, are
>> there any other differences between them?
> I'll take a stab at this because it's been on my mind lately... just
> my thoughts; not a definitive answer to your question.
> First of all METS is about expressing /structural metadata/ for
> digital objects -- usually complex digital objects consisting of
> different physical files that are somehow interconnected. As a
> practical matter it also provides a way to relate descriptive,
> technical, and administrative metadata to that digital object.
> METS is currently defined as a XML schema, but the only part of the
> schema that's required is the structural metadata (and the file
> inventory that goes with it). You can include the other kinds of
> metadata if you want, and they can be expressed in any other XML
> schema that has an associated name space.
> RDF isn't associated with any particular kind of metadata, but
> practically speaking it is usually expressed with its own XML
> RDFS, and is most often used for descriptive metadata (I think, but
> could be way off here).
> There were attempts in the past to encode structural metadata with
> (in Europe) which found that it added a lot of overhead to the
> processing of the metadata, which is about all you use structural
> metadata for (e.g. building a "page turner" display, or
> a website).
> That said, you could easily drop RDFS-encoded metadata for an object
> into a METS file, or you could convert a METS file into RDFS if you
> did a METS->RDFS mapping.
> The reason to consider that possibility is the tools... right now,
> seems easier to write custom tools for METS-encoded objects, or to
> METS as a simple transfer syntax between digital library
> In the future, as RDF tools and platforms mature, it might make
> to leverage them by coming up with an RDF schema for METS that we
> all live with and migrating our METS objects to that binding...
> shouldn't be hard at all. But the compelling reason to do that now
> yet to emerge.
> There's a project at Berkeley's Interactive University (led by
> Yee) which is looking at METS mapping to IMS and RSS, so he may also
> be thinking about the RDF-izing of METS-encoded digital objects and
> will hopefully speak up...
> I hope this helps you a bit,
> MacKenzie Smith
> Associate Director for Technology
> MIT Libraries
> Building 14S-308
> 77 Massachusetts Avenue
> Cambridge, MA 02139
> [log in to unmask]
Raymond Yee 44 Barrows Hall, #3810
Technology Architect UC Berkeley
Interactive University Project Berkeley, CA 94720-3810
[log in to unmask] 510-642-0476 (work)
http://iu.berkeley.edu/rdhyee 413-541-5683 (fax)