Notes on the METS Implementors Meeting
Chapel Hill June 15, 2006
Sponsored by the METS Editorial Board, the first METS Implementation Meeting was held on June 15, 2006 in as a post-meeting workshop in conjunction with the Joint Conference on Digital Libraries. Goals for the MIM were twofold:
- Provide an opportunity for representatives of METS implementations to meet and engage in technical discussions on issues related to the implementation and use of METS;
- Provide feedback to the METS Editorial Board regarding barriers to implementation and common problems that impede implementation so that the Board can prioritize strategic choices that would best facilitate the effective use of METS, and support the development of the METS community.
Participants and Workshop Format:
Workshop participants representing at least twenty institutions from around the world provided extended introductions to their institutions. These introductions plus the “Lightning Talks” contributed by a number of participants focused the topics to be discussed.
Participants recognized that the adoption of profiles was key to distributed use of METS. Consequently, the discussion identified barriers to the development and formal documentation of profiles, and generated ideas for facilitating profile development. Participants noted a reluctance to officially register a profile until it was well developed and tested internally. Formal registration implies a more formal and arduous approach to instituting changes in the profile, especially when used by others. Moving from a local environment to more distributed use is made more difficult by the need to replace local identifiers with the identifiers assigned once the profile is registered.
To facilitate the development and registration of profiles,
participants suggested the creation and maintenance of a public
“METS Profile Playground” to which METS community members
could contribute drafts of profiles, or sections of profiles
under development. Sections of the Profile Playground could
correlate to the various sections of the METS Profile Schema.
Participants expressed interest in exploring the modularization
of profiles to allow better development of profile subsections.
Another key consideration was the enhancement of the METS
website to allow better searching and browsing of the schemas,
the documentation and the example METS instances.
Of primary interest for the Tools discussion was identification of new and developing METS related tools as well as the needs for other tools. What ways could communication among tool developers be improved, and how could commercial tool developers become involved?
Lightening Talks revealed the existence of the following METS related tools
besides the ones already noted on the METS
website (Tools & Utilities):
an XSLT METS generation framework tool created by the California
Navigator, designed to deliver multi-page objects developed
by Indiana University (presentation
- ContentE (in Portuguese with English
translation), designed to support the construction of
formal structures of digital objects developed by the National
Library or Portugal;
- ResCarta tools, an open source suite of tools for building small
digital collections using METS, MIX and MODS;
- A collection of stylesheets created by the Library of
Congress to take existing HTML data from an integrated library
system catalog and a file system, and transform it into
a displayable, navigable XML structure (METS/MODS) presentation .
Discussion following the presentations generated the identification of other classes of tools needed such as tools encapsulating preservation related information. A suggestion was made to have a METS Tools event to which commercial and non-profile developers could be invited.
The operative question for this part of the discussion followed
naturally from the Tools discussion, i.e., is there a need
for a METS API and reference implementation? If so, what is
needed in a METS API that XML API’s cannot provide? Given
that Apache xmlbean and Harvard University’s METS
toolkit are available for Java, is there a need for similar
functionality in other languages?
The Lightening Talk provided background information on the questions, and
on a draft METS Beans application that has been created by
Markus Enders (SUB Göttingen) (presentation
and METS Beans attachments). From subsequent discussion,
it was clear that a METS API could be useful at some level
for the overall objectives of saving time in creating and
parsing METS, and avoiding the generation of invalid data.
Two levels for an API were defined: Level one was a means
to allow access to single METS elements, attributes and their
relationships. Level two would allow more complex functions,
such as creating instances of external schemas (MODS, PREMIS,
etc.) Participants thought that the second level, and a third
level (METS Profile aware) would be the most useful. They
identified barriers related to creation of specifications
for the API, actual code development, maintenance of the API,
and licensing, but seemed to think that the idea had great
merit and worth pursuing.
Participating institutions were using many of the endorsed external schema including MODS, MARCXML, Dublin Core, LOM, and DDI for descriptive metadata as well as MIX, textMD, AES, and the LC/AV elements for technical metadata. While many of the institutions were using METSRights for documenting access levels and rights owners, there was a general interest in making more examples visible with the various schema options available for rights, and for use of PREMIS for preservation metadata, as well as technical for a PDF application.
Suggestions for building a stronger METS community included improving communication, making more detailed technical documentation available along with examples of METS instances, and identification of who was using METS for what. Face to face meetings were encouraged, especially if they could be set up with different audiences in mind ranging from METS Beginners to Implementers at various stages of development. Implementers’ meetings would be especially valuable if participants brought actual situations for discussion (with and without presenters), and time were allowed for breakout sessions and problem solving.
Workshop participants were made aware of forthcoming technical
documentation for the METS schema being prepared by the METS
Editorial Board. The documentation is due out for public input
by September 30, 2006.