Evolution of a Processing System in Library*

a

Large Biomedical

BY LouISE DARLING, Librarian JAMES FAYOLLAT, Systems Programmer

Biomedical Library University of California, Los Angeles ABSTRACT The processing system used in the UCLA Biomedical Library is modest in size and still under development. Its origins date back to a batch mode serials control system begun in the mid-1960s. This was converted to an on-line system which currently has modules for check-in, updating and retrieval, claims. bindery preparation, and invoice information. Titles can be retrieved at the terminal by search of any word in the title, by subject heading, language, country of publication, and type of publication. The system is adaptable to network use and at present is shared with one other library. To the serials system has been added a computerassisted cataloging and card production system. The latter utilizes serials nucleus software as well as design for data input and data storage. In-house listings and coding procedures overlap in a general way. Work is under way on further integration of the two processing subsystems and a feasibility study has been completed for addition of a subsystem for acquisitions which will combine and adapt features of the other two; for example, information retrieval characteristics from both, catalog coding and programs for acceptance of data, serials programs for claims, and other output programs. Cost benet'its of' the subsystems are described and discussed.

THE still developing processing system used in the UCLA Biomedical Library had its origin in a batch mode serials control system begun in the mid-1960s. The library had reached a crisis point in managing its serials work load as the collection continued to expand rapidly while the staff size remained stationary. At that time we were not thinking in terms of a processing system but only of how best to master the serials situation. We were acquainted with the work on the University of California, San Diego campus * Presented June 4, 1975, at the Seventy-fourth Annual Meeting of the Medical Library Association, Cleveland, Ohio.

20

and the PHILSOM system at the Washington University School of Medicine Library in St. Louis and we had good prospects of support from our medical school for a venture into automation. As the project got under way all the now familiar problems cropped up, but, in spite of numerous mistakes, by 1968 a satisfactory batch serials system had replaced the manual system and the staff had developed at least a modicum of sophistication about applications of automation to library processing. The technology for on-line systems for libraries was becoming practical at about the same time and seemed to provide solutions for the less desirable features of the batch system such as time lag in availability of the most current information, accuracy and other problems in working with the card files required in batch processing, low place in the queue for computer time, the drudgery of keypunching records, and the difficulty in adapting to variant needs and unexpected situations. Because use of on-line techniques was then a new departure for libraries, we were able to secure a research grant from the National Library of Medicine (NLM) to develop the system and plunged in.' Success in this experiment led us to consider automation of cataloging and acquisitions by using online procedures. There was a possibility, never realized, that the University Research Library at UCLA might develop an acquisitions system, so in line with campus policy on duplication of efforts, we elected cataloging as the next area in which to work and are only now adding acquisitions although it would have been more logical to begin with the latter. The concept of a technical processing system began with the designing of the cataloging project. Before dis' National Library of Medicine Research Grant LM00739-92.

Bull. Med. Libr. Assoc. 64(l) Jan. 1976

PROCESSING IN A LARGE BIOMEDICAL LIBRARY

cussing the unifying features of the system, a brief description of the capabilities of what are now referred to as the three subsystems is in order. The serials control subsystem has been reported in detail by James Fayollat in three papers which appeared in the Journal of the American Society for Information Science during 1972-1973 and in other communications [1-4], but there have been important programming improvements and certain capabilities added in the meanwhile. The discussion which follows refers to current operations. In brief, the system is an interactive one permitting real time display of all parts of the record. It eliminates all card handling, provides for keyboard entry of all input, and through visual verification, promotes accuracy. Check-in of new serial arrivals and backfiles, updating and retrieval, claims, binding preparation, and invoice records are all handled at a CRT terminal. Programs are in PL/1 and designed for use on a time-sharing system. The terminal used by the Library is the Delta Data Systems model 5,000; two or more can be used simultaneously. There are now over 14,200 entries in the data base, 6,700 of them current, 6,100 ceased, 200 on order, and 1,200 cross-references. All the bibliographic fields of our record are compatible with the MARC II format. There are in addition many processing fields which are not included in MARC. To use the full MARC format would require more disk space and make manipulation of records more difficult. The data base is in sequential rather than random order because of variable length records which must be updated with high frequency and listings which must be processed in sequential order. Furthermore the records are stored very compactly; the average record contains 270 characters and an average of 44 records are contained on each track of the double density disk packs which are now used. Because of the high density storage the total disk space required for the records themselves and for the ancillary special purpose files is only 430 tracks. Tasks are monitored by 60 subroutines, about 20 of them frequently, the rest less often. Operations are at the 180K level but could be compressed to 130K or even 100K of memory space by greater use of overlay programming if the price schedule did not dictate otherwise. The on-line system is compatible with the Bull. Med. Libr. Assoc. 64(l) Jan. 1976

former batch system which is still used in modified form for listing purposes. As in the earlier system, the next expected issue is predicted for check-in; titles of irregular frequency which show only the latest issue added have a time interval assigned for automatic claiming. In designing the system, attention was centered first on maintaining and updating the data, second on retrieving information from it. Retrieval can be accomplished in three ways, the choice depending on the task at hand. The fastest and simplest method is to key in the seven digit accession number which appears on the lists, slips, or letters from which the operator is often working. The second and usual method for check-in and reference purposes is based on the inverted file concept. The file contains title words, subjects, languages, countries, and type of publication. Boolean logic can be used in searching where useful. A third approach used only for retrieval of titles with much-used words-science, national, biological, etc.-is by use of an arbitrarily constructed search code. In addition, there are special programs for retrieving information (usually listings) on vendors, frequency, and invoice records. Trial was made of public access via terminal in the reading room near the reference desk but there was strong preference on the part of both users and reference staff for a daily-now thrice weekly-print-out of complete holdings, claims, bindery, locations, call numbers and other information. This is easier and speedier to use than signing on at a terminal for the type of information generally sought about serials. The last five issues of the print-out are always retained so that there are enough copies to meet demand. A week's margin in currency is generally acceptable, but it must be admitted that there are times when there is an uncomfortable amount of competition for the latest issue. An audiotape installed at the print-out table gives clear, concise directions on how to read the print-out when reference assistance is not available. There are, of course, printed directions in every issue as well. The computer print-outs are supplemented by copies of the biennial published serials holdings list distributed in many places throughout the public areas of the library. The system is adaptable for a kind of network use but at present is shared with only one other library, the Medical Sciences Library 21

DARLING AND FAYOLLAT

Consortium on the University of California, Irvine campus. A recent account reports the comparative ease with which the UCLA work was adapted to the needs at Irvine [5]. Other libraries in Southern California have expressed interest in using the system on a network basis but do not at this time have active plans to do so. On the other hand, among the many people who have come in person to examine our operations, several have asked for the programs and more are waiting for the documentation package which James Fayollat, after a variety of unanticipated delays, is completing now. How many places have actually adopted any parts of the program we unfortunately do not know, but we assume few or none since very little has appeared in the literature about other on-line serials control installations. The one other place which has definitely declared its intention to use the system and has already begun implementation is the Lister Hill Library of the Health Sciences at the University of Alabama in Birmingham. The start-up costs for using the system are substantial because of the necessary conversion from manual to machine-readable form and the need to adapt the programs to the time-share operating program of whatever computer is to be used. Network use, of course, eliminates the latter factor. Ongoing costs, however, at least in our own case, have been decidedly costeffective in the areas of interlibrary loans, reference, bindery, and serials operations. Personnel in the latter, now that the developmental period is over, have dropped back to the 1964 level while time saved in the other areas has been transferred to other public service functions.2 Altogether we estimate that the system has saved us the equivalent of 2.5-3.0 library assistants: that is, $17,000-$20,000 a year, plus benefits impossible to obtain from the manual operation. Moreover the data base has increased by over 2,500 records since the start of the automation project. Since 1971 when ongoing costs were first established each record has been significantly augmented (such as beginning and ending dates, CODEN, ISSN, 2 The systems programmer has not been included in the costs because he occupies the slot formerly designated for Assistant Librarian for Technical Services. Salaries are not equal, but the programmer devotes only a fraction of his time to maintenance of ongoing systems, the rest to developmental work.

22

and place of publication), the number of subject headings has been increased severalfold and a title index to corporate entries has been added. The ongoing costs for rental and maintenance of two terminals and for computer time and related costs has remained at $14,000-$15,000 annually, with new program efficiencies balancing price increases. For example, when the Health Sciences Computing Facility made wholesale changes in its operating system in order to gain the advantages of the new double density disk packs, the library reformatted.the entire data base in order to make maximum use of the space. As a result disk space used by the library has been reduced approximately one-half, but new charges especially on paper, line connect time, and use of overlay techniques have cancelled the savings. Finally, it should be noted that the check-in assistant needs only two or three days of training to handle basic operations. The number of titles checked in per day averaged 150 with 25-40 the range per hour, depending on computer response time. The second subsystem of the technical processing system provides for computer-assisted cataloging. Details have been reported in a recent paper by Ruth Traister [6]. Programming is all in PL/1 as in the serials control system and the same make of terminal is used, but with a cassette attachment for temporarily storing data off-line. Current work load is approximately 600 titles per month of which 45% require original cataloging. For the rest 40% of copy is for titles in the National Library of Medicine Classification, and 15% for titles in the Library of Congress Classification. The prime objectives for the cataloging module were production of catalog cards sorted in filing order for eight catalogs and four special files, production of monthly acquisitions lists, and-for regional use-author/title and book catalogs. Demand bibliographies can also be produced for any categories identified in the fixed fields: authors, subjects, dates, languages, etc. but Boolean programs have not been developed because what interest there is in this capability is low and usually less expensively served by CATLINE searches. The usual way to retrieve information is by transaction number if it is at hand, otherwise by a search code modeled on that used by OCLC [7]*3 'Development of the cataloging module was supBull. Med. Libr. Assoc. 64(l) Jan. 1976

PROCESSING IN A LARGE BIOMEDICAL LIBRARY

Computer-related costs, terminal maintenance, and insurance average just under $6,000 annually, printing $3,600. The system has saved an appreciable amount of time, but in absolute terms produces no dollar savings compared with the manual system. On the other hand, there are important benefits: reduction in access to the latest information in the public catalog from six to two weeks or less, elimination of clerical chores from cataloger assignments, thereby freeing time for work on backlogs, the potential for absorbing larger work loads without staff increase and the several products available from the system for intra- and extramural use. The acquisitions module and the systems programming work which will accompany it should, however, in combination with the cataloging module produce at least minor dollar savings and a useful, flexible vehicle for all our technical processing work. It should probably be mentioned before going on to acquisitions that the library maintains a subject heading authority list which can be updated on the Delta Data CRT but is not at this time connected in any way with the cataloging subsystem. Should the latter eventually become a public use tool, it would be important to tie the subject heading list to the system. Based on a feasibility study made recently it will take an estimated eight to ten months to implement the acquisitions module. The acquisitions subsystem data base will consist of a merger of three files dealing with texts, monographs and sets: the in-process file (approximately 6,500 entries), the "hold" file (500-1,000 entries), and the approval file (300-400 entries). The in-process file includes books-on-order, books received and sent to cataloging, information slips and cross references to a separate standing order file for sets. The "hold" file is basically a second priority file for items which in this stringent budget period the library wishes to hold for critical reviews and/or to gauge user demand. It also contains items for which more information is needed before an order can be issued or items also in scope for other campus libraries and thus requiring cooperative decisions on who will buy what. The approval file is a record of citations for books expected from approval plan vendors. ported by National Library of Medicine Medical Library Resource Grant LM01263. Bull. Med. Libr. Assoc. 64(1) Jan. 1976

On-line merger of the files will accommodate the varieties of use and purposes by checktag mechanisms. The cataloging code sheet has been modified for joint cataloging and acquisitions use by addition of an acquisitions field. The search key formulation also follows the cataloging pattern, but there will be program modification to allow for searching by author and title routinely and, as in the serials system, by key words. Orders, claims, cancellations, conference and congress letters, expenditure and management information, and a variety of other listings including status lists for reference desk use will be available as well as the capability of creating records as may be required for campus and universitywide use. One of the more significant developments will be augmentation of programs from the cataloging system to accept acquisitions data, new programs to provide the interchange of data records between the acquisitions and cataloging systems, and programs to facilitate access to records of either data file. As already pointed out, the modules of our processing system use or will use identical hardware, on-line input, off-line output, PL/1 as the programming language and, most important, the same programmer who has worked in close cooperation over the years with a fairly stable library staff. These factors alone would be almost certain to bring common elements and unifying features into the three systems even if a conscious effort had not been made to work toward a single overall system. The serials system, being first in time, has become the basic unit. The cataloging system utilizes serials nucleus software as well as the serials design for data input and storage. In-house listings and coding procedures overlap in a general way. The acquisitions system will, with minor modifications, use serials programs for orders, claims, cancellation letters, management information, etc., catalog coding and cataloging programs for acceptance of data, and information retrieval characteristics from both programs. To make effective use of these interchanges and to manage the entire processing system efficiently, there must be additional integration into one multipurpose set of software which will handle all needs of all three systems. Programming still required, in addition to adding the necessary fields and options for

23

DARLING AND FAYOLLAT

acquisitions, consists primarily in providing indexes or other techniques for multiple use and multiple retrieval purposes. Individual records or parts of individual records will be described as being accessible through various combinations of indexes. A last step in sharpening the efficiency of the integrated system involves making comprehensive, although in most cases not extensive, modifications to existing software. Two types of modifications are planned. It will be necessary first to develop the software required to handle several terminals at once. This involves the programming concepts of multiprogramming and reentrant coding. When working properly, the same copy of program codes in the computer will direct the operation of two or more terminals as opposed to one copy for each terminal as is now the case with our system. The project will be experimental in some degree but the computing facility has told us that it is entirely feasible and agreed to assist and advise us because the work once done will result in reduced overhead on the computer. The second group of modifications center on integration of the nucleus software of this multiple terminal project with the technical processing system. Several months of fine tuning of the resulting system will probably be necessary to achieve the optimum cost/performance configuration. The major result should be a significantly reduced overhead in the computer because one terminal management program will now take the place of as many programs as there are terminals operating simultaneously. This reduced overhead will, of course, be reflected in reduced computer charges to the library. This technical processing system has served the library well thus far and should serve even better when the acquisitions module is completed. It has many sophistications which are likely to be some time in coming to us via regional or national processing networks, and at this point it is not feasible economically or politically to consider going over to a minicom-

24

puter system. We are, of course, only too well aware of the intermittent problems inherent in depending on a research computer center on the one hand and on the other the impossibility of a single institution system such as ours developing a favorable cost ratio in comparison with the cost ratio of multiuser regional and national systems. We have considered the advisability of promoting a small local network operating through the HSCF on our programs, but in the light of current developments this would probably be practical only for serials control where as yet no other systems are up and where we know from the Irvine adaptation that implementation costs are relatively modest. For the long haul, though, we reluctantly lean toward the recent prediction by Richard De Gennaro that individual, localized systems efforts will disappear 181. It is in keeping with the times. REFERENCES 1. FAYOLLAT, JAMES. On-line serials control in a large biomedical library: Part I, Description of the system. J. Am. Soc. Inf. Sci. 23:318-322, Sept.Oct. 1972. 2. FAYOLLAT, JAMES. On-line serials control in a large biomedical library: Part II, Evaluation of retrieval features. J. Am. Soc. Inf. Sci. 23:353-358, Nov.-Dec. 1972. 3. FAYOLLAT, JAMES. On-line serials control in a large biomedical library: Part III, Comparison of online and batch operations and cost analysis. J. Am. Soc. Inf. Sci. 24:80-86, March-Apr. 1973. 4. FAYOLLAT, JAMES. On-line serials control at UCLA. In: Proceedings of the Clinic on Library Applications of Data Processing, 1972. Urbana, Illinois, University of Illinois Press, 1972. p. 69-81. 5. LOEPPRICH, JOEYCE. Implementing an on-line serials control system. Paper presented June 4, 1975, at the Seventy-fourth Annual Meeting of the Medical Library Association, Cleveland. Ohio. 14 p. 6. TRAISTER, RUTH. Computer-assisted cataloging: experiences at the UCLA Biomedical Library. Bull. Med. Libr. Assoc. 63:283-294, July 1975. 7. KILCGOUR, FREDERICK G. The shared catalo,ing system of the Ohio College Library Center. J. Libr. Automation 5: 157-183, Jan. 1972. 8. DE GENNARO, RICHARD. Library automation: the second decade. J. Libr. Automation 8:3-4, March 1975.

Buoll. Med. Libr. Assoc. 64(l) Jan.

1976

Evolution of a processing system in a large biomedical library.

The processing system used in the UCLA Biomedical Library is modest in size and still under development. Its origins date back to a batch mode serials...
798KB Sizes 0 Downloads 0 Views