Journal of Library Metadata, 9:134���152, 2009 Copyright �� Taylor & Francis Group, LLC ISSN: 1938-6389 print / 1937-5034 online DOI: 10.1080/19386380903095123 Encoded Archival Description (EAD) Metadata Scheme: An Analysis of Use of the EAD Headers BRIAN CARPENTER American Philosophical Society, Philadelphia, Pennsylvania, USA JUNG-RAN PARK College of Information Science and Technology, Drexel University, Philadelphia, Pennsylvania, USA An exploratory study was designed to gain a perspective on the cur- rent state of Encoded Archival Description (EAD) metadata use and quality. This article examines the implementation of EAD meta- data at 6 digital archival repositories. It identifies EAD metadata elements that evince the most frequent inconsistent and incomplete usage. For this, the header eadheader areas of a collection of 150 randomly collected EAD metadata records were analyzed in terms of frequency, completeness, and consistency. In addition to these criteria, 3 best practices guidelines for EAD were consulted to determine how EAD metadata tags are used in local repositories vis- ` a-vis the guidelines. Analysis focuses on patterns of usage of certain aspects of EAD metadata within each repository as well as compar- ison across all 6 repositories. Implications and issues drawn from the evaluation are also discussed in relation to the issue of semantic interoperability across EAD digital repositories. KEYWORDS Encoded Archival Description, EAD header, meta- data, metadata quality evaluation, best practices guidelines, digital repositories INTRODUCTION The development of the Encoded Archival Description (EAD) metadata scheme began in 1993 under the guidance of a project team based at the Address correspondence to Brian Carpenter, American Philosophical Society, 105 S. 5th St., Philadelphia, PA 19104, USA. E-mail: bcarpenter@amphilsoc.org 134

Encoded Archival Description (EAD) Metadata Scheme 135 University of California at Berkeley Library. The Berkeley team, headed by Daniel Pitti and made up of archivists and librarians from several institutions, set out to construct an encoding standard that would make available a range of information on archival materials that were not possible within the struc- ture of MARC. Owing to its flat structure, MARC is limited in fully describing a hierarchical finding aid (Chang, 2001). Given the unique qualities of these materials, such as the quantity and diversity of items that populate individual archival collections and the pri- macy of finding aids that serve to facilitate discovery within them, the group articulated primary requirements around which they would structure Docu- ment Type Definitions (DTDs) for use as finding aids. DTDs define elements and attributes expressed in the Extensible Markup Language (XML) docu- ment. The EAD DTD is a standard for encoding archival finding aids using XML. To the best knowledge of the authors, there are no extant studies evaluating EAD metadata records. We speculate that this is in part due to the complexity of the EAD DTD and of the archival finding aids. The EAD DTD consists of a total of 146 element tags such a large set of data elements presents challenges in assessing metadata quality among EAD implementers. Tracking the use of such a large set of elements is inevitably beyond the purview of this study. Recognizing that these different tags perform a vast array of distinct functions and are used in varying circumstances in individual repositories, the scope of this study was narrowed to an examination of a smaller set of tags that occur in common situations and are amenable to analysis in roughly the same way. We will therefore focus only on the header eadheader areas of the sample EAD records. The EAD header contains the overall title for the archival collection that the finding aid covers, along with essential adminis- trative metadata on the creation of the EAD record itself. This information includes the author of the EAD record, the date of its creation, the repository publishing the record, the language in which it is published, and the history of any revisions that have been made to the EAD record. As the EAD header covers information that is common to all archival collections, this section of the EAD record can be compared uniformly record-to-record. For these reasons, EAD headers are best suited to a general analysis of EAD usage at different repositories. The goal of this exploratory study is to examine how EAD header metadata elements are being used across six digital archival repositories and to identify EAD header metadata elements that evince the most fre- quent inconsistent and incomplete EAD metadata application. This study also aims at examining the effectiveness of the metadata guidelines in improving the quality of metadata. Also discussed are implications drawn from the evaluation of the current status of the EAD metadata application

136 B. Carpenter and J. Park in relation to the issue of semantic interoperability across digital archival repositories. The goals of this study are directly derived from the following research questions: 1. How are EAD header metadata elements being used across the surveyed digital archival repositories? 2. How do the metadata guidelines and best practices function in the quality of metadata? EAD STRUCTURE: AN OVERVIEW The EAD scheme manifests the hierarchical nature of archival description. Given their hierarchical structure, finding aids can be over 100 pages long and may surpass the amount of data that a MARC record can contain (Chang, 2001). The EAD DTD consists of a total of 146 element tags that fall under four top-level elements. These elements can be split evenly into two primary areas. The first contains summary information on the finding aid itself, cov- ered by the Header eadheader and Frontmatter frontmatter elements. The second records summary description of the contents of the archival materials themselves, covered by the Archival Description archdesc and Description of Subordinate Components dsc (Pitti, 1999). The eadheader fulfills the need for the most basic of publication in- formation access points and general administrative information on the find- ing aid. It also provides standardization in the inclusion and sequencing of this information across all EAD-encoded finding aids. There are four main sub-elements in the eadheader . The first of these sub-elements, the EAD Identifier eadid , provides a unique identifier for the finding aid and may also contain a code denoting additional information such as the location, type, or source of the archival materials. The eadid is followed at the same level by the File Description filedesc subelement. The filedesc itself subdivides into five lower levels that together cover the basic bibliographic description of the finding aid, including title, subtitle, author, sponsor, publication place and date, and edition and series information (where applicable). The filedesc sub-element is followed by Profile Description profiledesc , which records the language of the finding aid, the author of the EAD version of the finding aid, and the date on which it was encoded. These first three sub-elements of the eadheader are, by definition, ap- plicable to all finding aids. The final Revision Description revisiondesc sub-element summarizes any changes that have been made to the finding aid, if any (Ruth, 1997).