random quote Link: Publications Forum Link: About DLF Link: News
Link: Digital Collections Link: Digital Production Link: Digital Preservation Link: Use, users, and user support Link: Build: Digital Library Architectures, Systems, and Tools
photo of books






Please send the DLF Executive Director your comments or suggestions.

Towards a Shared Cataloguing Tool for VR Collections

Planning for the DLF's Academic Image Cooperative (AIC) demonstrated that a shared, web-accessible cataloging database comprising rich descriptions of art historical works promises to be of enormous value to educational and cultural communities. Like comparable bibliographic utilities (OCLC, RLIN), it would serve both scholars and catalogers. For scholars and students it would act as invaluable reference tool. For custodians of visual resource collections in libraries and other institutions, the database would inform and facilitate local cataloguing activities; begin to minimize the redundant effort that image cataloging activities presently involve, by providing an authoritative, critical mass of cataloging records available for local repurposing; and ultimately foster appropriate local implementation of standard controlled vocabularies. The database, finally, could have a number of potential business applications, for example, to those active in the private market for art historical objects.

Although a prominent feature on the want lists of appropriate professional communities, the construction of such a database has so far proved too complicated to try.[1]

  • High-quality descriptive information exists in machine-readable form but is created with very different metadata standards and encoding schemes.
  • Harmonization, if feasible, is itself problematic. In stark contrast to bibliographic records, it is impossible, indeed undesirable, to agree a single authoritative description for an art historical work. Attributes change over time (architectural structures are renovated; paintings are cleaned) and descriptions are contingent upon authors' perspectives and upon the users' information requirements.

Yet recent developments afford an unparalleled opportunity to successfully assemble the database described above.

  • Professional consensus has emerged around the Core Categories for describing art historical works as proposed by the Visual Resources Association. The Core, until recently an evolving concept, has stabilized to a large extent with the recent publication of version 3.0. Because the Core is not proposed as a prescriptive standard, it supplies a common framework within which it may be possible to present a single, unifying view of fundamentally heterogeneous descriptive data. Because it has emerged from within a professional community that is exclusively concerned with visual resources, the common ground it proposes is richer than that supplied by other metadata standards such as the Dublin Core. At the same time, the visual resources community appears to be more accepting of the notion that its descriptive information is likely to be more informative and more useful if it is developed and presented with variance in tact, rather than shoe-horned into constraining and prescriptive metadata schemes.
  • SGML, and now XML, provide syntaxes capable of representing multiple views of a single record. Although the capability is well known there has until recently been little implementation experience of it within the educational and cultural communities. At present, it is possible to point to relevant work at a number of sites which demonstrates capability in robust operational service environments.
  • Software applications capable of supporting a genuinely distributed and dynamic approach to collection development - one that relies upon data development at two or more participating sites - are more well developed today than they have been in the past. OCLC's CORC, the University of Washington's Content software, and version 3.0 of Luna Imaging's Insight software offer dynamic collection development tools that are operated in a web-based environment.[2]
  • Through its work developing and demonstrating a prototype image service, the AIC is in receipt of several offers in principle to supply machine-readable records describing the works of art that are housed in a number of very large visual resource collections. In addition, it has received expressions of interest from the Universities of Michigan and Washington, and from Luna Imaging, Inc., to participate in various capacities in processing any records that are supplied and to make them available through some uniform view in a web-based utility.

The AIC accordingly proposes a prototyping initiative that will:

Assemble a prototype database of between 300,000 and 500,000 descriptive records sourced from between 3 and 6 institutions. Records will be supplied in their native formats and mapped by the AIC to a specified implementation of the VRA Core. Once assembled, the database will be evaluated by a small group comprising scholars and visual resource professionals. The DLF and the AIC will also evaluate technical, organizational, and business implications involved in transitioning the prototype into operational service.

Although more detailed functional specifications for the prototype will be developed at an initial stakeholders' meeting, the following service description is offered as a starting point for discussion and derives. It derives from a meeting of the AIC held in New Haven on 3-4 August 2000.

The prototype will at a minimum:

  • Integrate between 300,000 and 500,000 descriptive records sourced from different machine-readable catalogues. Records will be mapped by the AIC to a uniform implementation of the VRA Core Categories version 3.0 but also available in their native formats.
  • The database will supply web-based browsing, searching (both fielded and keyword searching), and retrieval of sorted result sets and it will be possible to display records within a result in brief and full modes.
  • The application will incorporate controlled vocabularies to enhance searching and to aid visual resource professionals in cataloguing. Introduction of controlled vocabularies will be phased and the prototype is likely to begin with a single vocabulary such as Union List of Artists' Names, Art and Architecture Thesaurus, ICONCLASS, Library of Congress Subject Headings, Thesaurus of Geographical Names.
  • The database will supply tools to enable bona fide "trusted" users to edit existing records and add new ones. Such users will be able to amend existing records and to relate individual records to one another either because they refer to the same work or because they exist in a parent-child relationship. Trusted users will also be able to assign key terms derived from implemented controlled vocabularies to a record or an element within a record. The database will attribute all amendments to the authoring individual.
  • Resource export ("copy cataloging"). To support local cataloguing efforts, users will be able to export records (as amended) in some common format for importation to and use in their local catalogue databases.

The project is expected to take place over the course of 2001 and to run in tandem with a test distribution of the Academic Image Cooperative service. It will begin by identifying stakeholding participants including data suppliers, data users, and applications developers, and representative domain specialists. Stakeholders will be convened in a meeting to develop a detailed functional specification for the prototype and its evaluation as well as and key milestones for the project's progress. Thereafter, the prototype will be built according to specification and subjected to formal evaluation and review by project participants.


1. The REACH and VISION projects played a significant role in the promulgation of metadata standards for art objects and provided valuable data about the complexities and challenges of creating and sharing image- and object-based information across communities and systems. The REACH project attempted to extract information of interest to researchers from commonly-used museum management systems, and to assemble that information into a single database with a Web interface. Overtaken by other developments, including the tandem evolution of the VISION project, the project was terminated. REACH differed from the current prototype in at least one fundamental respect insofar as it attempted to integrate information about objects in a heterogeneous group of museums. The collection scope of the current prototype is more rigorously defined by fine art and architecture, and further circumscribed by virtue of the prototype's focus on image collections created in support of overlapping and in many cases common academic curricula. The VISION project was similarly constrained to art historical images from academic visual resources collections and other research collection but also less ambitious than the current prototype. Fundamentally it set out to evaluate application of the VRA Core Categories. The project is documented in VRA Bulletin 25:4(winter 1998). VISION and REACH are compared in Elisa Lanzi's contribution to Art Documentation 17:1(1998).

2. See http://www.oclc.org/corc/, http://content.engr.washington.edu/, and http://www.luna-imaging.com/insight.html.

return to top >>