Automatic identification of "works" toward construction of frbrized opacs: An experiment on JAPAN/MARC bibliographic records

Purpose: Efforts have been made to improve the OPACs by collocating bibliographic records sharing the same "work" and navigating users among records under a certain "work," according to the Functional Requirements for Bibliographic Records (FRBR). This paper investigates methods of automatically identifying "works," i.e., grouping bibliographic records sharing the same work, for the JAPAN/MARC records, which are typical Japanese bibliographic records created and maintained by libraries in Japan. It reports the extent to which records can be automatically identified as members of a particular work and also which of the possible methods are effective. Methods: The method used in this study is to generate work identification keys for each work represented in a bibliographic record and then to bring the keys representing the same works together. The keys are in principle constructed as a combination of an author name and a title from the record. Several methods of generating such keys were examined and the clustering of keys was executed for each method. The clusters built automatically were evaluated by comparing them with the sample correct sets built manually. Results: The results of the experiment show that the proposed method is effective in average cases; however, the performance depends on the characteristics of works, for example, the volume of records sharing the same work, whether anonymous or not, and whether uniform titles exist. It also shows that it is effective to generate keys for every bibliographic hierarchical level with data elements such as author headings, statements of responsibility, descriptive titles, and title headings.