Greetings,

On behalf of the Digital Library Working Group, University Libraries,
University of Minnesota, I am forwarding some comments about the proposed
METS Standard. These comments were written by Charles F. Thomas, Digital
Projects Coordinator at the University of Minnesota, and then reviewed by
the Digital Library Working Group.

I will send our comments embedded as text below, and attach a MS Word file
with the same information.

Thank you for the opportunity to comment on the METS standard.

Josephine Crawford, chair
Digital Library Working Group
University Libraries
University of Minnesota


                                * * * * * * * * * *


A Response to the METS Proposed Standard
From the University of Minnesota Libraries
Digital Library Working Group
05/21/02

The Metadata Encoding Transmission Standard (METS) offers an extensible and
scalable model for organizing metadata about digital objects. Its ancestral
roots in the Making of America project are apparent; METS offers a rational
solution for managing large bodies of individual electronic files that are
closely interrelated. In particular, we see METS as a vital precursor to
next-generation presentation and navigation tools for digital objects such
as digitized books and other information-bearing-objects that consist of
many discrete parts. METS is flexible in that it permits any xml-based
metadata schema to be employed within its "wrapper" structure. The creators
of METS also have incorporated features that may be of benefit in other
functions such as digital archiving.

We do offer two constructive criticisms for the Digital Library Federation.
The first is that METS is based entirely upon an XML Schema Description
(XSD). This choice is understandable, XML Schema are themselves XML-based
and hold the promise of easier data and data structure interchange.
However, we would suggest that a Document Type Definition (DTD) version of
the METS standard co-exist with the XSD statement of its structure. XML
Schema undoubtedly offer much, but a large body of middle-of-the-road
pragmatists will hesitate to adopt XML Schema at this time, because of its
recent vintage and perceived instability. DTDs, conversely, are much more
established as a way to define structure for encoding. If the XSD version
of METS took advantage of some of the advantages XML Schema offer, such as
data-typing, then this would not be possible. However, since METS is
designed to be a simple, flexible wrapper for metadata (and potentially
even digital object content), a co-existing DTD-based statement of the METS
structure and rules should promote faster adoption among the XML
communities who still prefer DTDs.

Our second comment is closely related to the first. The METS overview and
tutorials frequently mention the capability of METS to actually wrap
Base-64 binary content into the metadata descriptions, for such purposes as
digital archiving. Strategies for permanently associating metadata with
digital objects have continued at least since the earliest discussion that
led to the Dublin Core metadata standard. To date, we have witnessed
multiple approaches to hard and soft associations between digital objects
and their metadata. One common way of doing this is through defining both
metadata and digital objects as related "entities." Defining entities is a
useful way of modularizing object and metadata management. Unfortunately,
XML Schema do not support the definition of entities such as text files,
binary objects, etc. This is one main reason that the XML Schema has never
purported to replace DTDs, which do permit the definition of entities.

The University of Minnesota Libraries Digital Library Working Group
appreciates the opportunity to comment on the proposed METS standard. It
demonstrates leadership by proposing a flexible and extensible means of
storing, referencing and interchanging metadata. We feel that stewardship
and maintenance of METS by the Library of Congress also is a very positive
sign. Our few concerns and suggestions deal mainly with accommodation of a
large body of existing institutions and practices, and the ability to
manage metadata and objects effectively. We support METS as a standardized
means of exchanging information, and hope that our suggestions will prove
useful.