Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=wjlm20
Download by: [University of Chicago Library] Date: 10 November 2015, At: 12:55
Journal of Library Metadata
ISSN: 1938-6389 (Print) 1937-5034 (Online) Journal homepage: http://www.tandfonline.com/loi/wjlm20
Providing Metadata for Compound Digital Objects:
Strategic Planning for an Institution's First Use of
METS, MODS, and MIX
Michael Dulock & Christopher Cronin
To cite this article: Michael Dulock & Christopher Cronin (2009) Providing Metadata for
Compound Digital Objects: Strategic Planning for an Institution's First Use of METS, MODS, and
MIX, Journal of Library Metadata, 9:3-4, 289-304, DOI: 10.1080/19386380903405199
To link to this article: http://dx.doi.org/10.1080/19386380903405199
Published online: 10 Dec 2009.
Submit your article to this journal
Article views: 263
View related articles
Journal of Library Metadata, 9:289–304, 2009
Copyright © Taylor & Francis Group, LLC
ISSN: 1938-6389 print / 1937-5034 online
DOI: 10.1080/19386380903405199
Providing Metadata for Compound Digital
Objects: Strategic Planning for an Institution’s
First Use of METS, MODS, and MIX
MICHAEL DULOCK
University of Colorado at Boulder Libraries, Boulder, Colorado, USA
CHRISTOPHER CRONIN
University of Chicago Library, Chicago, Illinois, USA
The University of Colorado at Boulder recently engaged in a grant-
funded pilot project to use Metadata Encoding & Transmission Stan-
dard (METS), Metadata Object Description Standard (MODS), and
NISO Metadata for Images in XML Schema (MIX) for a collection
of digitized Sanborn fire insurance maps of the state. This article
will draw on this experience to outline the processes and decision
making required to implement new metadata structures, and will
offer some insights on planning strategically for an institution’s first
use of these increasingly important metadata standards.
KEYWORDS metadata standards, compound digital objects,
strategic planning, METS, MODS, MIX
INTRODUCTION
Sanborn maps are historical fire insurance maps of towns and cities at the
detailed level of individual blocks, streets, and buildings. Beginning in the
1860s, the maps were created by the Sanborn Map Company to aid in assess-
ing the risk of fire damage and, therefore, in setting insurance premiums. To-
day, the historical maps are valuable resources for the study of architecture,
urban planning and development, demographics, environmental conditions,
and even genealogy.
In 2006, the University of Colorado at Boulder Map Library received an
Institute of Museum and Library Services (IMLS) grant through the Colorado
Address correspondence to Michael Dulock, University of Colorado at Boulder Libraries,
Cataloging & Metadata Services Department, 184 UCB, 1720 Pleasant Street, Boulder, CO
80309-0184, USA. E-mail: michael.dulock@colorado.edu
289
Downloaded by [University of Chicago Library] at 12:55 10 November 2015
290 M. Dulock and C. Cronin
State Library under the provisions of the Library Services and Technology Act
(LSTA) to digitize a portion of the maps that were no longer under copyright
restriction. The segment of the collection selected comprised 346 maps on
2,385 sheets, detailing 79 cities in 52 counties in Colorado covering the years
1883–1922.
Sanborn maps consist of multiple sheets that together form the entire
map, with an “index sheet” showing the relationships between each sheet in
the map. Within the context of the maps selected for this project, the num-
ber of sheets per map ranged from one to 98. In a digital environment these
types of resources are often referred to as compound (or complex) digital
objects, wherein the object itself comprises multiple discrete components
that are usually organized in a specific way to form a cohesive whole. The
compound structure of the map sheets, combined with a goal of standard-
izing technical metadata capture, resulted in the project team’s decision to
explore implementing Metadata Encoding & Transmission Standard (METS,
available at http://www.loc.gov/standards/mets/) as part of the digitization
plan for the collection.
One of METS’ important strengths is its ability to support metadata pro-
vision for compound digital objects beyond just descriptive metadata. In
contrast to other metadata schemes—like MARC, Dublin Core, VRA Core,
and EAD—METS facilitates the association, organization, and collocation of
technical, structural, descriptive, provenance, and behavioral metadata for
digital objects within a single METS wrapper. To this end, METS addresses
some of the challenges inherent in the distributed environment in which
digital objects—and the metadata that describe them—are managed by the
organization, as well as found and used by patrons. METS packages together,
or points to, all the information about a resource’s archival master, access,
preview, and metadata files, thereby reducing reliance on the institutional
memory of a few to recall the various locations of those files. In think-
ing ahead to possible future scenarios, METS’ standardization will allow the
University Libraries to access, reuse, recondition, or repackage digital ob-
jects and metadata (e.g., migrating to different delivery or storage systems,
preservation reformatting of digital objects, integrating collections with those
of other institutions).
When the project team was originally exploring METS, there were few
commercially or freely available METS creation tools in the marketplace. The
project team explored the development of a homegrown tool that generates
METS records from existing but dispersed metadata sources. Digitizing the
Sanborns also provided an opportunity to explore the use of an established
standard for technical metadata: the Metadata for Digital Still Images in XML
(MIX) standard, maintained by the Library of Congress and NISO. Following
this standard allows for consistent recording of specific details on the process
of digital imaging, including equipment and software u sed for digitization.
This article will discuss how the team developed various metadata creation
Downloaded by [University of Chicago Library] at 12:55 10 November 2015
Providing Metadata for Compound Digital Objects 291
tools and how it built technical metadata capture into the automated pro-
cess of digital imaging, as well as some of the lessons learned in how to
plan strategically for an institution’s first implementation of Metadata Encod-
ing & Transmission Standard (METS), Metadata Object Description Standard
(MODS), and NISO Metadata for Images in XML Schema (MIX).
CASE STUDY
The process of evaluating whether and how to implement METS revealed
that the personnel involved did not possess all of the skills required to build
an application that could create METS records. Team members from Digital
Resources Cataloging had knowledge of the metadata schemes and standards
involved (XML, METS, MODS, MIX, MARC, MARCXML); representatives from
the Map Library had knowledge of both the Sanborn maps and their users; the
Digital Initiatives Librarian had scripting experience and expertise with the
University Libraries’ digital asset management system, Luna Insight. No one,
however, had the programming capabilities required to build an application
that could generate METS records.
The team allocated a portion of the grant funds to hire a contractor to
develop a METS generation application. The amount of funding available was
limited, as was the timeframe for the grant, so there was a desire to work with
a programmer who was local. The team pursued hiring a graduate student
from the Computer Science program on campus. A qualified doctoral student
was identified, with whom the team met to discuss deliverables, available
funds, and timelines. Including the student programmer, the project team
totaled six individuals.
Requirements for the functionality of the tool were created collabora-
tively according to the needs of all the departments involved. Staff from
the Map Library addressed patron needs and collection concerns. Digital
Resources Cataloging provided input on descriptive and technical metadata
standards, as well as application functionality from the perspective of cata-
logers who would actually use the tool. The Digital Initiatives Librarian pro-
vided direction related to local information technology systems and sought to
ensure a broad understanding of how various facets of the project would, or
would not, integrate with existing University Libraries infrastructures. These
requirements were discussed at the first meeting of the entire group and
were readdressed as needed throughout the duration of the project.
Because METS, MODS, and MIX were new to everyone on the team,
taking time to educate ourselves on the standards was an important part
of the process. A group-study approach was adopted to facilitate broad
understanding of what all aspects of the project would entail. While the
catalogers on the team took primary responsibility for learning about and
communicating the implications of adopting MODS for descriptive data, the
Downloaded by [University of Chicago Library] at 12:55 10 November 2015
292 M. Dulock and C. Cronin
process of learning about both MIX and METS were shared by everyone
in the group. To ensure that whatever tools were developed fit all of our
needs, the Computer Science student also familiarized himself with all METS
documentation, and joined discussion and planning meetings. It was decided
that all documentation for the process, the product, and the METS packages
specifications, would be written collaboratively.
With the requisite skills in place for the creation of a METS generation
application, the project team assessed what metadata already existed, and
how it could be repurposed to create the desired METS records. Following
the University Libraries’ policy to create separate records for different formats,
electronic resource MARC records, containing descriptive metadata for each
digitized map, were derived from the print records in OCLC.
File-naming conventions were devised at the outset of planning, both for
the process of developing the METS tool and for communicating deliverables
to the digitization vendor. Continuity of file naming was essential so that each
subordinate metadata unit would be uniquely identifiable and matched with
the correct image file for the map. Base file names were constructed using
the first three letters of the city and the last two digits of the year represented
by the map (e.g., a map of Erie in 1893 became “eri93”). While there was
discussion of including all four digits of the year (in order to avoid conflicts
across centuries), it was felt that the digital collection would probably not
expand beyond 1923, making such conflicts unlikely.
Base file names were added to a local 9XX field in each MARC record;
the METS application was designed to use the contents of this field to auto-
matically establish file names for the METS records (e.g., eri93.xml), as well as
names for the constituent parts, or sections, of each METS file. For example,
the MODS descriptive metadata section for Erie 1893 would be “DMDeri93,”
and the administrative metadata section would be “AMDeri93.” The vendor
also provided image filenames for each sheet according to these conventions
(e.g., “eri93001,” “eri93002,” and so forth) as well as the images’ correspond-
ing MIX technical metadata files, (e.g., “TMDeri93001,” “TMDeri93002,” and
so forth). A truncated outline of the METS XML structure for the ‘Erie 1893”
example appears in Figure 1.
The basic framework for the METS records was based on examples avail-
able at the METS Web site. While the Sanborn maps constitute compound
digital objects, they do not exhibit some of the more complex features that
more complicated METS implementations have exploited (i.e., multiple lan-
guages, parallel files, etc.). Thus, a model was created for the application that
includes a simple structural map, descriptive metadata for each object, and
technical metadata for each component image. The <structMap> section of
the METS record, which defines the hierarchical order of the compound ob-
ject for presentation, uses the ORDER attribute along with filenames to repre-
sent the relatively simple, “flat” nature of the collection. Ninety-five percent of
the map sheets follow simple numerical order, so that the lowest-numbered
Downloaded by [University of Chicago Library] at 12:55 10 November 2015
