At 07:34 AM 4/2/2002 -0600, Peter C. Gorman wrote:
>At 1:30 PM -0500 4/1/02, Julia Flanders wrote:
>>I hope this doesn't seem like a betrayal or a deep heresy (my turn to
>>get the flames!) but I've heard good things about the possibility of
>>using the METS schema, which allows you to bring together metadata,
>>text transcription (e.g. in TEI), and multiple images (whether of the
>>original text or of associated images) and specify their behaviour. I
>>don't know a whole lot about it but it looks pretty elegant; they're
>>starting to use it at the Brown digital library projects. Those who
>>actually know something about METS should speak up and correct or
>>amplify what I've said.
>
>
>On the other hand, one can store both the page image references and
>the marked up text in the same (TEI) file, and use TEI as the DIP for
>both methods of access. While METS is certainly fine to the task, it
>may be superfluous if you already have a TEI delivery infrastructure.
>A great strength of METS, however, is the way it allows you to
>associate descriptive and administrative/technical metadata with
>objects at any level of granularity (I've often wished there was a
>direct way to attach descriptive metadata at the TEI DIVn level).
>
>Still, even e-facsimiles of print books are essentially textual
>structures (even if skeletal), and textual structures are what TEI
>does best. The structural metadata maps fairly straightforwardly:
>
>http://www.library.wisc.edu:4000/dept/ltg/DigiLib/EFacs/EFacsDataDictionary.html
>
>Near the top, there's a link to a document mapping the data elements
>to TEI structures. We're in the process of sewing together the e-facs
>and 'real' TEI interfaces in order to drive both from the same source
>file.

Yikes, this is going way outside of the boundaries of the TEI!  I've
included some references for those who wish to learn more.

Julia referenced METS (Metadata Encoding & Transmission Standard).  You can
find out more about METS at:
http://www.loc.gov/standards/mets

Peter mentioned DIP.  Before you go off to P3/P4, etc., this is not a TEI
element, it's the Dissemination Information Package, which is part of the
OAIS (that's Open Archival Information System).  You could either find out
more about OAIS than you ever wanted to know by going to:
http://ssdoo.gsfc.nasa.gov/nost/isoas/
*or* you could read a pretty nice summary in Appendix A of this paper:
http://www.rlg.org/longterm/attributes01.pdf

Okay, the original question was how to store metadata related to
images.  It depends what type of metadata.  If it's all sorts of hairy
technical metadata that records how the image was created, I would argue
that is outside of the bounds of the TEI.  If you want to link the page
image to the transcription, I'd look at the approach used at Michigan and
elsewhere (extending pb to include more information than just n, so you can
record the page number and it's sequence).  Peter is right, TEI used in
this manner is a perfectly good means of dissemination if you have a
transcription and scanned pages.  If you have technical metadata about each
and every one of those scanned images, and would like to store that
information in an archivally sound way, I think it makes sense to store it
outside of the TEI.  At this point, the TEI does not support this type of
metadata and you would be on your own in extending the TEI to make it
fit.  There is a lot of good work going on in defining best practice for
technical metadata for still images and audio-visual.  METS provides a file
inventory and a structural map (which is based, very loosely, on TEI divs),
and "buckets" for various other types of metadata.  You can provide links
between your METS document (where the md lives) and your TEI document
(where the transcription lives) very easily.  METS is not a
one-size-fits-all solution, but neither is the TEI.

Best,

Merrilee

p.s. John Price Wilkin gave me a variety of TEI-encoded documents (ranging
from raw to cooked) from Michigan's Making of America site, and I have
fallen down on my promise to make METS examples that worked with these
documents.  I'll try to do so in the near-term future.