About the presenter:

Caroline Arms has been at the Library of Congress since 1995, as a technical program coordinator for the National Digital Library Program based in the Information Technology Services division of the Library. In particular, she has been the technical advisor for the Library of Congress / Ameritech National Digital Library Competition. Between 1997 and 1999 this competition made awards for twenty-three projects to digitize primary source materials to complement and enrich the Library's American Memory resource. By October 2000, twelve have been integrated into American Memory. Prior to joining the Library, Arms worked at the Falk Library of the Health Sciences at the University of Pittsburgh, as the first Director of Computing at the Amos Tuck School of Business Administration at Dartmouth College, and providing computing support to researchers at the University of Sussex and the Open University (in the United Kingdom). She has a B.A. in Mathematics from Oxford University and an M.B.A. from Dartmouth College. In the late 1980s, Arms edited two volumes for EDUCOM, Campus Networking Strategies and Campus Strategies for Libraries and Electronic Information, both published by Digital Press.

Summary:

The Internet has stimulated the development and deployment of collections of digital content managed and made available over the network for particular communities or purposes. These digital libraries, with their associated services, have varied ancestry. Some, like American Memory have been built by libraries or other archival institutions. Others have emerged from user communities to provide shared management and networked access for important digital resources, such as survey data for social scientists, sensor data from satellites or telescopes for astrophysicists and other scientists, or instructional resources for faculty and teachers.

The metadata elements needed to allow specialist users to find, identify, select, and obtain the resources they need and to navigate the web of relationships among them do not necessarily match the elements and rules for bibliographic cataloging of materials traditionally held by libraries. The potential for coordinated access to resources of different types from different sources, however, calls for a level of commonality among metadata schemes. Simple and rapid access to full content may reduce the need for some cataloging details, since the user may be able to use the full content or an automatically created summary, such as a thumbnail of an image or outline derived from marked-up text, to aid selection. On the other hand, although archival collections in paper form are often described as a whole or at the level of a series or physical container, item-level identification is essential in a digital library, increasing the cataloging cost. However, content in digital form can be a source for automatically generated metadata; such metadata will be less costly but flaws that would be easily corrected or avoided by a human cataloger may go undetected. In digital libraries, not all relationships between items have to be recorded in catalog records. Relationships between digital works can be embedded when the work is created or derived automatically by analysis of the full content. Citations can link to the works referenced, providing navigation capabilities far richer than those possible through catalog records.

This paper will draw on experience gathering together metadata from heterogeneous sources for American Memory, particularly for the collections digitized and cataloged at other institutions through the LC/Ameritech competition. It will also reflect on several initiatives to develop rich structured metadata schemes for specific domains and others to find simple approaches to support resource discovery across domains. Trends and commonalities will be identified and influences among metadata schemes highlighted.