Publishing Bibliographic Data On Optical Disks: A Prototypical Application And Its Implications

Abstract

Bibliographic databases are quite large and relatively static, making them ideally suited for storage on optical media now becoming available. Consequently, libraries and publishing houses are pioneering certain applications of optical disks. The high storage density and the economy of optical disks make them attractive media for large databases. However, two factors create problems in the design of sophisticated retrieval software: certain device characteristics (e.g., slow access time compared to magnetic media), and the user interface requirements for bibliographic data. We discuss a prototypical application of optical disks to the publishing of the Library of Congress's machine-readable collection (the MARC and REMARC databases). The prototype disk is manufactured from pre-mastered videotape and is controlled by an IBM PC microcomputer. The prototype represents a significant effort in system integration, requiring development of hardware interface, error correction, low-level access software, high-level language interface, and user-friendly front-end applications. To prepare the data for indexing and display, mainframe software was developed. Based on our experience with the prototype, we discuss a number of issues common to the publication of such databases: data structures, search strategies, precomputation of retrieval results (a powerful technique well-suited to optical disk applications), and issues pertaining to standardization and publication formats.