Academic journal article Information Technology and Libraries

The Expert Cataloging Assistant Project at the National Library of Medicine

Academic journal article Information Technology and Libraries

The Expert Cataloging Assistant Project at the National Library of Medicine

Article excerpt

Over the last few years, the National Library of Medicine (NLM) has pursued the development of an expert system to assist catalogers with personal name authority work. This article reviews the project, from project and tool selection through knowledge engineering to evaluation, and discusses the reasons for deciding to redirect NLM's energies from development of an operational system to modification of the cataloging rules.

During 1985 and 1986 staff from the National Library of Medicine, the Library of Congress, and the National Agricultural Library met to discuss possibilities and plans for expert systems development in the area of technical services. NLM's Library Operations (LO) staff was interested in taking advantage of this promising new technology to enhance cataloging productivity and/or efficiency. Cataloging was selected since it is such an expensive activity and therefore a potential area for cost savings, and it requires both knowledge of a complex set of rules and their professional-level interpretation.

After much planning and discussion with staff in the NLM's research and development branch, the Lister Hill National Center for Biomedical Communications (LHC), a position for Systems Librarian in the Office of the Chief of the Technical Services Division (TSD) was established primarily, but not exclusively, to be the system developer for this project. As a librarian who had been a cataloger for three and a half years and who had taken courses on expert systems in library school, I was selected for this position in July 1989. Other NLM staff also have been involved in the project, including the chief and deputy chief of the division, for planning and oversight; two cataloging liaisons for their perspectives; and a computer scientist in Lister Hill for technical consultation.

Planning the Project

The initial project objective was to use artificial intelligence and expert systems technology to reduce the level of human effort required in the costly cataloging process. Although TSD expected eventually to cover the whole range of cataloging, a manageable portion was needed for the initial stage of the project. Authority work was chosen because it is one of the most time-consuming, and therefore costliest, aspects of cataloging. (An in-house study conducted in 1986 showed that for monographs cataloging, NLM spent 45% of its cataloging time on authority work, compared with 36% on descriptive cataloging and 19% on subject cataloging and classification.) Authority work also requires significant training time for new catalogers. Expert systems technology is usually applied to activities where either a small amount of time can be saved on a short, but often repeated process or a large amount of time can be saved on a longer process. The project team chose to work on an activity of the first type, selecting personal names, which are the most frequently occurring name headings in NLM's catalog records.

The system was to be designed to assist the human cataloger in selecting the form of a personal name heading to be used in catalog records. The program would take the form of the name found in the bibliographic description, search NLM's and LC's name authority files, examine the search results, and download an LC authority record if appropriate. If a match was not found, the program would recommend to the cataloger what the correct form of the name should be according to the Anglo-American Cataloguing Rules, 2nd ed. (AACR2) and create the local authority record.

The project team planned for the initial knowledge base of the system to contain rules from AACR2 (especially chapter 22, "Headings for Persons"), the Library of Congress's Rule Interpretations (LCRI) for those AACR2 rules, local NLM practice, communications protocols for accessing NLM's and LC's computers, and search syntax for the name authority files. Then knowledge engineering techniques would be used to extract from professional catalogers information on how to construct authority file search statements based on the form of the name in the description, how to interpret search results, and how to apply the rules in the aforementioned tools. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.