All,

A while back, Jerry encouraged some of us to get together, talk about "best
practice" for METS, and report back what we found out. Well, we haven't
discovered a lot, but we are working towards figuring it out together. We
have decided that it's too early for Best Practice Guidelines, (upper
case), so we are aiming for first practices, or lower case bpg. Here are
my minutes from a meeting spurred by geographic fortune. There is a lot of
work to be done, and we are hoping that others will pitch in. We will
report back from time to time.

Best,

Merrilee

Notes for METS "best practice guidelines" meeting, 7 May 2002

Attending:
Stanford: Nancy Hoebelheirich, Michael Olson,
Berkeley: Lynne Grigsby-Standfill (UCB Library), Guenter Waibel (Berkeley
Art Museum/Pacific Film Archive)
RLG: Merrilee Proffitt, Ricky Erway, Robin Dale, Fae Hamilton, Judy
Gerstle, Mike Sanderson, Anne Van Camp

Briefly noted: Bruce Washburn, Bill Washington, Judith Bush, Linda West,
Susan Yoder

Goal of the meeting: to see if any or all of the represented institutions
(Berkeley, Stanford, RLG) have shared approaches, and to see if there is
enough commonality between these organizations to begin work on first
practice documents.

Following introductions, each institution discussed their current and
perspective future use of METS.

RLG: "owns" one collection (Marriage, Women, and the Law), in the position
of converting collections from a multiplicity of sources for use in
services. Right now, the focus is on using METS to support display of
complex objects. RLG is also trying to look forward to having data
contributors give more than just derivative surrogate files and descriptive
metadata in order to create METS objects suitable for digital
archiving. Currently, RLG is using METS for data rescue or data triage and
would like to move beyond this. Have experimented with producing different
viewers for different objects. Struggling with the idea of creating two
(or more?) types of METS documents, one for service delivery and one for
archiving. The beauty and the danger of using links (METS document
pointing out) and the beauty and danger of wrapping all metadata and files
(very large files, difficult to update, impossible to serve out). This was
a theme that was touched on by all meeting participants, and clearly and
area for more exploration and discussion. RLG (and others, it sounds like)
may not be using METS directly for viewers, but instead decomposing METS
documents for use by viewer software.

Stanford: currently working with one very large collection (2.5 millions
pages, mostly scanned page images, also includes photos). Struggling with
bundling the data files that represent the collection in a large METS
object, with EAD serving as a hub document with pointers into the METS
document for lower-level information. Focus is more on developing best
practice than on developing tools. Idea of having different ratings for
collections, with some treated at a very low level (strucMap, fileInv, some
additional minimal metadata for functionality), others being prepared for
digital archiving (with a richly defined AIP).

Berkeley: lots of MOA2 documents, converting to METS in the near future. A
range of projects, from producing core METS document (structMap + fileInv)
to link scanned table of contents and index pages to MARC records, to more
complex objects (scanned pages, transcription files, audio files). Working
on a variety of tools, that produce MOA2 objects, a MOA2 viewer all to be
adapted to METS soon. CDL has a pending grant proposal out to create a
digital repository for all nine campuses. Using METS for everything,
packaging metadata even with single data files. Undertaking data rescue on
previously digitized collections (no technical metadata or variable
technical metadata). Berkeley Art Museum/Pacific Film Archive has a tool
that produces XML as part of imaging workflow: EAD, MOA2, TEI. Want to be
able to export out to desired formats, in order to contribute to consortial
projects such as RLG Cultural Materials. Concerns with evolving extension
schemas. For audio and video, when is it best to use SMIL, MPEG7, when is
it best to use METS? Guenter chairs a group under the OAC/CDL that is
grappling with issues of interest to this group, and agreed to share
documentation as it is developed. This group agreed to give feedback to
Guenter.

Discussion of future steps and goals:

There seemed to be plenty of work that could be done, both by members of
this group and by others. Some potential deliverables identified by the group:

Decision tree/roadmap document
Best practice for fileInv/StructMap
Extension schemas
         Required elements, mapping
Viewer specs
Discussion of: pointing vs. embedding; link management; data update and
version control
Training/education

Of these, the most desirable/easily accomplished seemed to be development
of a "Road Map" document, which would help guide new users or those
approaching the standard with some useful advice about what they would need
to know or think about before approaching METS. Merrilee agreed to oversee
development of this document.

The second task that the group decided to take on was development of best
practice for Native METS, or the fileInv/structMap for non-time based
materials, (i.e. page turned, potentially hierarchically structured
documents, book-like objects). Lynne agreed to take on this, given enough
example documents (everyone agreed to pitch in and share examples of these
components with her).

Nancy agreed to develop a third document, a decision tree for evaluating
available extensions schemas for descriptive metadata (Dublin Core, MODS,
etc.).

None of these documents is seen as being an end-all, be-all, and are meant
to stimulate community discussion and eventually understanding of working
best practice. Each document would be made available for wide review and
discussion, and could be revised or re-written as needed.

Other tasks were viewed as larger, more complicated, and more appropriate
for the larger community. RLG has an interest in viewer development, and
in sharing approaches with others, and could potentially host a small
meeting for those actively engaged in viewer development.


Merrilee Proffitt
Research Libraries Group -- www.rlg.org
1200 Villa Street, Mountain View, CA 94041 USA
voice: +1-650-691-2309 -- [log in to unmask]