of the American Society for Information Science

Go to
 Bulletin Index

Volume 25, No. 6

bookstore2Go to the ASIS Bookstore

August / September 1999



Theory and Practice in the Organization of Images and
Other Visuo-Spatial Data for Retrieval

by Corinne Jörgensen,
guest editor of special section of the Bulletin

With the rapid evolution of tools enabling the creation of collections of networked digital images, image indexing and description have now become of critical interest to a wider community. There are, however, inherent difficulties in this process (long recognized by image librarians and archivists). The essential problem is that neither images nor their users are homogeneous. Images, furthermore, are quite likely to be used in contexts widely differing from their original functionality. Image access thus requires levels of description that can accommodate various types of images and image attributes, users, collections and tasks.

The four articles presented here all describe different attempts at addressing the "image problem." In these articles, image description is discussed variously as indexing, cataloging, classifying or assigning metadata, but the essential problem of developing a succinct and usable surrogate for the image remains – a surrogate that can represent both the image content and its potential usefulness. Reflecting the diversity of images, the papers reflect a range of approaches from philosophical musings to practical application and from theoretical to research oriented investigations.

James Turner's paper starts at a fundamental level of developing a framework within which to think about the different areas related to visual collections. The collection as a physical thing requires a different type and level of information than the individual images within a collection. Turner's typology reveals many important activities associated with visual collections. The variety of institutions dealing with visual collections, professions which use these visual collections and production/consultation methods for images all carry implications for their description. Enumerating these various facets should help to ensure that important areas for image description are not overlooked.

The O'Connor and O'Connor article focuses on the potential users of image collections, presenting verbal descriptions of images by "naïve" users and contrasting these with classifications or descriptors from subject headings lists. They add an additional factor to be considered besides image content or topic, that of functionality of the image. Thus, they add to the classification process the "question state" of the user as well, noting that different types of uses imply a need for different types of image description. Further discussion explores the need for indexing of interpretive or affective attributes and raises a number of interesting questions, as well as suggesting new ways of handling image indexing in a more dynamic way.

The last two articles deal with specific types of collections with strong visual components: a historic costume collection and a collection of geospatial information objects. One interesting aspect is the common need among these materials for not simply a basic level of description but a rich representation of the "documents," a need shared by text-based materials. These two projects also highlight the fact that access to such visual materials is, however, still primarily text-based.

Goodrum and Martin describe a costume collection that is seemingly ideal for inclusion of visual surrogates. Inclusion of such surrogates, however, has been problematic in other collections. Copyright restrictions often prevent their inclusion, and even where visual materials are freely available, system constraints such as storage often intervene. The authors discuss the interesting problem of developing a classification system for a diverse collection for which no current system is adequate. While the classification of this visual collection remains text-based, the discussion highlights an urgent need for visual cataloging aids, a need shared by many other visual collections being catalogued by non-experts.

Fraser and Gluck describe a metadata study which demonstrates that visual or geospatial document metadata share the same problems as other text-based indexing, that of usability. Similar studies have been conducted in LIS on back-of-the-book indexes, highlighting the role that formatting and display can make in the user's ability to perceive (and thus use) indexing information, whether it is presented in printed text or on a computer screen. Their article underscores the fact that basic research in human-computer interaction and human factors must not be forgotten in the rush to provide expanded access through computers and brings the emphasis upon the user back into the system.

Taken as a whole, we can view the articles as a microcosm representing the complex nature of image collections and their participants and the many facets which need to be considered when trying to manage these collections for the widest possible access.

Corinne Jörgensen is assistant professor in the School of Information and Library Studies at the University of New York, Buffalo. She can be reached there by mail at 534 Baldy Hall, Buffalo, NY 14620-1020; by phone at 315/476-5205; or by e-mail at


How to Order

@ 1999, American Society for Information Science