Academic journal article Information Technology and Libraries

The Next Generation of Public Access Information Retrieval Systems for Research Libraries: Lessons from Ten Years of the MELVYL System

Academic journal article Information Technology and Libraries

The Next Generation of Public Access Information Retrieval Systems for Research Libraries: Lessons from Ten Years of the MELVYL System

Article excerpt

This paper views the design of the next generation of public access information retrieval (IR) systems in higher education from the perspective of a decade of development, deployment, and operation of the MELVYL online system at the University of California (UC). It highlights design decisions and assumptions that were made for the MELVYL system that have proved advantageous, as well as those that have proved limiting or have led to dead ends. Our design choices were probably similar to those made by most other online catalog designers at the time. Some decisions at UC that have proved in hindsight to be shortsighted or cowardly (and also a few that proved better than we might have hoped) were only guesswork, because there was no base of experience from which to work. Other decisions were artifacts of limited functionality and capability from the underlying base of information technology upon which the catalog was built, or of a limited budget to acquire resources. Particularly in the case of computing hardware, it was not that desired technology did not exist ten years ago (unlike certain supercomputing applications--visualization being the most striking example--that emerged during the 1980s), but that the cost of the desired computing cycles, memory, and mass storage was out of reach. Costs of these resources have dropped now sufficiently that they can be used more freely as we consider systems for the 1990s.

The available base of software technology was a different matter. The limited functionality in the software components, such as database management systems (DBMS), that might be used to build an online catalog was a serious problem. In 1980, the DBMS choices were few, and none of them was entirely satisfactory. Interestingly, as we consider future directions for the MELVYL system in 1992, the choices seem to have improved little in terms of functionality, although the available commercial software has matured considerably in terms of stability and performance. The full set of functionality still seems tantalizingly out of reach, manifested most broadly in database systems that remain as research vehicles within the computer science research community, and thus unsuitable for production use in a system the scale of the MELVYL catalog.

Finally, in terms of delivery platforms, we viewed the system as limited by the installed base of character mode ASCII terminals and so designed to the lowest common denominator "glass teletype." In theory, we might have procured a special terminal for use with the MELVYL system (as some other systems had done), since, as discussed in more detail later, our initial assumption was that most terminals for catalog access would be placed in libraries. But we felt it was important to be able to support the installed base, presuming that networking on the UC campuses would continue to improve and that over time more of this installed base would be able to reach the catalog. Given the explosion of networking that occurred later in the 1980s, this proved to be a very wise decision as it greatly facilitated wide access to the catalog.

The history and current status of the MELVYL system has been amply covered in the papers that have appeared in the two previous "MELVYL at Ten" special sections of Information Technology and Libraries and in the spring 1992 issue of the DLA Bulletin. But a review of the design assumptions and system objectives for the original MELVYL online catalog, many of which, to my knowledge, were never explicitly articulated and debated as part of the planning process prior to its development, forms an essential part of the context for this paper. Thus, the first part of the paper reviews them with the benefit of ten years of hindsight, along with certain realities of the information technology base of the late 1970s. The remainder of the paper focuses on key problems that emerged as we gained experience with patron use of online catalogs at UC and elsewhere, and as the MELVYL system has grown larger, more complex, and more capable. …

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

Oops!

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.