“The Charles W. Cushman Collection: Enhancing Visual Resource Discovery Through Descriptive Metadata Based on Subjective Image Analysis”

Linda Cantara Indiana University Libraries lcantara@indiana.edu

The Charles W. Cushman Collection° comprises nearly 18,000 Kodachrome slides, captured from 1938 to 1969 by Charles Weever Cushman, an accomplished amateur photographer, world traveler, and pioneer in the use of color photography. Cushman bequeathed the collection to his alma mater, Indiana University, on his death in 1972, yet its existence was virtually unknown until late 1999 when a university archivist rediscovered the suitcases in which the slides were stored, along with Cushman’s notebooks containing identifying information and dates for each slide. The images visually document in color the vernacular history of people, places, and events that have previously been seen only or primarily in black and white. To provide online access to the collection, a project team at Indiana—including staff of the Digital Library Program, the University Archives, and the University Libraries— has digitized the slides and has designed a relational database to document administrative, technical, and descriptive metadata about the images. Transcriptions of Cushman’s notebook annotations and corresponding documentation on slide mounts will be keyword searchable. In addition, we are creating descriptive metadata for each image by assigning terms for subject content, genre, physical characteristics, and geographic location. The terms are selected from standard controlled vocabularies, primarily the Library of Congress Thesauri for Graphic Materials (TGM I and TGM II), but also Library of Congress Authorities Files (LCAF) and the Getty Thesaurus of Geographic Names (TGN). When all the slides have been cataloged, the database will be mapped to an XML format—most likely Encoded Archival Description (EAD) but possibly Metadata Object Description Schema (MODS)—and will be searchable over the Web. The use of controlled vocabularies in the descriptive metadata of a large image collection has the potential to optimize visual resource discovery, but many issues must be addressed and resolved to ensure the expense of assigning the terms truly results in improved image retrieval:

The Library of Congress Thesaurus for Graphic Materials I: Subject Terms (TGM I) and Thesaurus for Graphic Materials II: Genre and Physical Characteristic Terms (TGM II) include topic headings developed by the Library of Congress Prints and Photographs Division over the past fifty years to catalog graphic materials. In adherence with ANSI/NISO Z39.19-1994,° terms are expressed in natural language word order following American English spelling conventions. Although new terms are added regularly based on proposals submitted by catalogers (Alexander), these thesauri frequently lack the specificity desirable for describing individual images. In addition, regardless of the domain knowledge and cataloging expertise of personnel performing image analysis to assign subject headings, inconsistencies are inevitable. Professional catalogers of traditional media readily acknowledge that subject analysis of printed materials may vary from cataloger to cataloger as well as from day to day. In this paper, I will discuss how we have attempted to approximate consistency of image analysis by multiple catalogers, the quality control procedures we have implemented to further ensure uniformity of subject term assignment, and the “work- arounds” we have devised to deal with encountered inadequacies in the selected controlled vocabularies.
The Digital Library Federation, the Getty Grant Program, and the Andrew W. Mellon Foundation in cooperation with Rice University are currently sponsoring a Visual Resources Association (VRA) review and evaluation of existing data content standards and current practice in order to compile a guide to good practice for describing cultural objects and images. The forthcoming Cataloguing Cultural Objects: A Guide to Describing Cultural Objects and their Images—the CCO Guide—will provide recommendations and examples for using data value standards tools and building authority records.° Used in concert with XML data structure and communication standards, the CCO Guide will facilitate consistent creation of descriptive metadata for visual resources, enabling interactive and complex search options. Paradoxically, Open Archive Initiative Metadata Harvester (OAI-PMH) service providers require data providers map rich metadata to unqualified Dublin Core, while content aggregators like the Research Libraries Group (RLG) Cultural Materials service must currently treat all controlled vocabulary terms as keywords.° Although neither OAI service providers nor RLG preclude the use of multiple or complex metadata formats, content creators nevertheless face a quandary: why commit extensive and expensive personnel hours to create increasingly rich metadata when interoperable and federated access is possible when implementing simpler, less costly, metadata standards?°
The foremost concern when designing and implementing an image retrieval system should be the searching needs of the user community (Sundt). Art historians, for example, may be trained in the use of one or more controlled vocabularies specific to art images,° but users of vernacular photography collections may have limited experience with any controlled vocabulary. Since lack of familiarity with the underlying vocabulary tools may actually diminish rather than enhance user interaction, the search interface should ideally facilitate direct mining of the encoded metadata, directing the user to broader, narrower, related and cross-referenced subject terms. This paper will conclude with a review of our usability test findings in respect to a prototype interface that integrates TGM I and II terms into the search facilities for the Cushman Collection.

REFERENCES

Arden Alexander Tracy Meehleib. “The Thesaurus for Graphic Materials: Its History, Use, and Future.” Cataloging & Classification Quarterly. 2001. 31: 189-211.

. Encoded Archival Description. : ,

. Getty Thesaurus of Geographic Names (TGN). : ,

. Guidelines for Implementing DC in XML. : ,

. Library of Congress Authorities. : ,

. Library of Congress Thesauri of Graphic Materials: TGM I (Subjects) and TGM II (Genre and Physical Characteristics Terms). : ,

. Metadata Object Description Schema (MODS). : ,

. Open Archives Initiative FAQ. : ,

Christine L. Sundt. “The Image User and the Search for Images.” Introduction to Art Image Access: Issues, Tools, Standards, Strategies. Ed. Murtha Baca. Los Angeles: Getty Research Institute, 2002. 67-85.