The value of MARC data, or, challenges of frbrisation

Bibliographic records should now be used in innovative end-user applications that enable users to learn about, discover and exploit available content, and this information should be interpreted and reused also beyond the library domain. New conceptual models such as FRBR offer the foundation for such developments. The main motivation for this research is to contribute to the adoption of the FRBR model in future bibliographic standards and systems, by analysing limitations in existing bibliographic information and looking for short- and long-term solutions that can improve the data quality in terms of expressing the FRBR model.

MARC records in three collections (BIBSYS catalogue, Slovenian National Bibliography and BTJ catalogue) were first analysed by looking at statistics of field and subfield usage to determine common patterns that express FRBR. Based on this, different rules for interpreting the information were developed. Finally typical problems/errors found in MARC records were analysed.

Different types of FRBR entity-relationship structures that typically can be found in bibliographic records are identified. Problems related to interpreting these from bibliographic records are analyzed. Frbrisation of consistent and complete MARC records is relatively successful, particularly if all entities are systematically described and relationships among them are clearly indicated.

Advanced matching was not used for clustering of identical entities.

Cataloguing guidelines are proposed to enable better frbrisation of MARC records in the interim period, before new formats are developed and implemented.

This is the first in depth analysis of manifestations embodying several expressions and of works and agents as subjects.