OCLC explores the implications of what could be the biggest change to library catalogs in more than a century

By Tom Storey
In 1998, the International Federation of Library Associations and Institutions (IFLA) announced a bold proposal to streamline cataloging and simplify resource discovery for library users—The Functional Requirements for Bibliographic Records (FRBR).

FRBR is an “entity-relationship model” of metadata for information objects, rather than a single, flat record conceptualization underlying current cataloging standards. It distinguishes among:

• A work—the distinct intellectual or artistic creation;
• An expression—the intellectual or artistic realization of a work;
• A manifestation—the physical embodiment of an expression of a work; and
• An item—a single copy of a manifestation.

Today, library catalogs are composed of bibliographic records that describe manifestations—books, manuscripts, periodicals, videos and other formats. Bibliographic units are described out of context.

FRBR restructures catalog databases based on particular works rather than on the various forms in which the works are expressed. It groups all versions of a given work together using a high-level record that links to numerous lower-level records, thereby collapsing near-duplicate items into a single entry point. With the FRBR model, each item must be described in context and related to the other items comprising the work.

Manipulating information using this approach aggregates what can be an overwhelming number of record retrievals into a manageable set that matches a user’s search terms.

“FRBR has the potential to improve cataloging, discovery and delivery by bringing versions of works together, showing relationships of various kinds, and enabling users to navigate to their level of interest,” says Thom Hickey, OCLC Chief Scientist.

For example, a high-level entry record for Humphry Clinker, which today has 184 records in WorldCat, would describe this intellectual creation of Tobias Smollett with links to foreign language translations, large print editions, and the numerous editions for which new illustrations, introductions, notes and bibliographies have been added over the years.

What does FRBR mean to libraries? How difficult will it be to convert existing catalogs into “FRBRized” databases? Will it help libraries reduce costs and library users find what they want more quickly?

In late 2001, OCLC researchers Rick Bennett, Jean Godby, Thom Hickey, Brian Lavoie, Ed O’Neill, Diane Vizine-Goetz and Jenny Toves began a series of experiments to explore the implications of FRBR and to look into the practical difficulties in implementing its approach. They used WorldCat, the world’s largest library catalog, as the test bed.

Preliminary findings suggest:

• The FRBR notion of a work is a valuable concept. It provides a means to aggregate bibliographic units to simplify database organization and retrieval.
• Works can be reliably identified from bibliographic records.
• Bibliographic records do not contain sufficient information to reliably identify expressions. Reliable identification requires physical examination of selected items.
• Applying FRBR may not be as burdensome as first thought. Research suggests that 80 percent of the bibliographic records in library catalogs represent single manifestations of works. Conversion efforts therefore could be focused on the 20 percent that represent works with multiple manifestations, where users would benefit the most from FRBR conversion.
• Developing computer algorithms to automate at least part of FRBR conversions would be beneficial. Manual conversion of existing files is likely to be prohibitively expensive.
• FRBR may potentially reduce the cost and increase the quality of both original and copy cataloging. Bibliographic information, particularly subject headings and classification numbers, used in a work can be “inherited” by all manifestations associated with the particular work, alleviating what can be a time-consuming and expensive process.

OCLC releases FRBR algorithm

OCLC is making an algorithm available free of charge to organizations interested in converting their bibliographic databases to the FRBR model.

Developed by OCLC Research, the FRBR algorithm makes it possible for users to write computer programs to generate sets of records that can be grouped for display as single works, making it easier for information seekers to find what they are looking for. The algorithm is available from the OCLC Research site at http://www.oclc.org/research/software/frbr/.

OCLC’s FRBR algorithm describes an automated process that extracts information from MARC21 records, compares it with a standard name authority file, and then brings the records together, based primarily on their authors and titles.

For example, in OCLC’s WorldCat database there are records for more than 400 different forms of Arthur Conan Doyle’s The Adventures of Sherlock Holmes. Using the FRBR algorithm, those records can be brought together as a single work, making it easier for librarians and other information seekers to identify the most appropriate items to meet their specific needs.

The algorithm is efficient enough to be run on large databases, such as WorldCat. It is also meant to be understandable, so that a librarian creating a record will be able to predict what other records it will be associated with, and understandable for patrons so they are not surprised by the groupings.

FRBRizing WorldCat

OCLC plans to use the FRBR model as it implements WorldCat’s new database technology to enhance the user’s experience. Based on a random sample of 1,000 bibliographic records gathered in December 2001, here are some statistical estimates as to how FRBR impacts WorldCat.

• 47 million bibliographic records (as of 12/2001) can be traced back to 32 million works
• The average work in WorldCat has approximately 1.5 manifestations
• More than 25 million—nearly 80 percent—of the 32 million works consist of a single manifestation
• Ninety-nine percent of all works in WorldCat have seven manifestations or less; 1 percent of all works have eight or more manifestations
• About 30,000 works have more than 20 manifestations; approximately 400,000 works have five or more manifestations
• The most widely held work in the sample had 1,251 manifestations and total holdings of 27,434
• The most widely held work in the sample that had just one manifestation has total holdings of 710

OCLC Research on FRBR

FRBR Work-Set Algorithm

Chief Scientist Thom Hickey led the development of computer algorithms to automate FRBR conversions.

Web site: http://www.oclc.org/research/projects/frbr/algorithm.htm
Publication: http://www.dlib.org/dlib/september02/hickey/09hickey.html

FictionFinder: A FRBR prototype for fiction in WorldCat

Research Scientist Diane Vizine-Goetz uses the FRBR model in a prototype system for searching and browsing 2.5 million bibliographic records representing fiction.

Web site: http://www.oclc.org/research/projects/frbr/fictionfinder.htm
Publication: http://staff.oclc.org/~vizine/CNI/OCLCFRBR_files/frame.htm

A Case Study: The FRBRization of Humphry Clinker

Research Scientist Ed O’Neill uses Tobias Smollett’s The Expedition of Humphry Clinker to determine if the information available in the bibliographic records is sufficient to reliably identify FRBR entities.

Web site: http://www.oclc.org/research/projects/frbr/clinker/default.htm
Publication: http://www.oclc.org/research/publications/archive/2002/oneill_frbr22.pdf

FRBR: Extending the Case of Humphry Clinker

In a follow-up to the initial case study, conversion algorithms are tested against a sample of 1,000 works from WorldCat.

Web site: http://www.oclc.org/research/projects/frbr/works.htm
Publication: http://www.oclc.org/research/publications/archive/2003/lavoie_frbr.pdf

CAPCON becomes OCLC CAPCON Service Center | Interview: Barbara B. Tillett