College and Research Libraries AUDREY N. GROSCH Implications of On -line Systems Techniques for a Decentralized Research Library System On-line and batch mode computer-based library systems are defined to show their basic differences. This is related to how on-line systems can provide an operating framework for the large research library, thus providing the groundwork for a national information netw01'k. This article also discusses systems design factors, operating system and application software, various geographic implications, user serv- ices, and library staff effects. Additionally, certain problems are noted which libraries face as they proceed toward design, debugging, and installation of computer-based on-line operating systems for record and information control and dissem,ination. ON-LINE AND BATCH MODE SYSTEMS PREDOMINANT among the computer- based library systems representing the current state-of-the-art are systems op- erating in the batch mode. In such systems, input data are gathered for processing at stated time intervals, and required outputs are produced by the · processing of these data. These outputs are usually printed or punched. Correc- tions, additions, or changes must wait until the next processing time. For many tasks, such as order printing, circulation card punching, and various listing func- tions, this mode has provided sufficient- ly current service on an economical ba- sis without detriment to user services. Certain library-user-related functions do suffer, however, because of the time lag between processing, correction, and Mrs. ·Grosch is Systems Coordinator in the University of Minnesota Library. 112 I desired output when batch mode is em- ployed. Functions such as recording re- ceipt of periodicals, charge-out of bibli- ographic materials, file searching file correcting, and timeliness of information are directly affected by the batch mode. It is highly desirable to perform these functions within the framework of an on-line computer-based information sys- tem. Such a system would permit a re- turn to the "real-time" environment which does exist when making changes in manual systems, yet provide all op- erating units with access to the total record base of the library system without file duplication. Operating in an on-line mode would mean that updating of files, initiation of transactions of various types, and ce1tain output functions would be processed in what appears to be «real-time" to the system user through the direct man/ machine interface of a console unit. Such a library system can be designed today for the large, decen- tralized research library since hardware Decentralized Research Library System I 113 and operating system software are avail- able for such systems. Project B.allots1 and work carried out at the University of California, Berkeley, under Professor Maron2 show that progress in this di- rection is being made. In addition, our national libraries are devising standard record formats for bibliographic informa- tion as typified by Project MARC. This on-line library system would be developed to serve library user and staff needs by creating a centralized, integra- ted data store of the library's biblio- graphic and in-process information. Such a system would operate via key- board display consoles, data collection units, and other necessary equipment connected to a central computer over various types of communication lines. Certain tasks would be pedormed in the batch mode, but main emphasis of the system to serve current user needs would structure the system to require short user response time, let us say, un- der four-second maximum delay on a message sent in response to a console inquiry. Therefore, the on-line, integra- ted record system envisioned by this author in future large libraries will com- bine both modes of operation for maxi- mum efficiency, service, and economy. Such a record system will be compatible to standard bibliographic information disseminated through our three national libraries, the Library of Congress, the National Library of Medicine, and the National Agricultural Library. 1 Stanford University. Bibliographic automation of large library operations using a time sharing system (Project Ballots), a proposal for research and/ or re- lated activities submitted to the U.S. Commissioner of Education for support through authorization of the Bureau of Research. Principal Investigator: Allen Barnet Veaner. Project Director: Rutherford David Rogers. May 1967, granted October 1968. 2 University of California, Berkeley. A study of the organization of search of bibliographic holdings rec- ords in on-line computer systems, a proposal for r€:. search and related activities submitted to the U.S. Commissioner of Education for support through authori- zation of the Bureau of Research. Principal Investi- gator: M. E. Maron. May 1967. SYSTEM DESIGN FACTORS No library system exists which truly integrates all of the total library opera- tions from the original order of an item to its eventual delivery to the user. Cer- tain of these segments are being per- formed by computer-based systems in various ways by many institutions. To achieve maximum benefit from an on- line application it is highly desirable to interlace as many operations and tasks as possible, i.e., to have the system per- form as much work as possible ~ within limits of the hardware procured. This serves to create orderliness in process- ing, efficiency on the part of the staff, and economical equipment utilization. For example, present circulation control systems of either operational mode re- cord the necessary transaction informa- tion, but the machine readable input card used to record the transaction in most of these systems is not a by-prod- uct of the original book processing op- eration, nor are these circulation files accessible by .any other means than tele- phoning the circulation department, consulting their file listing, or using their console for file inquiry. Further, the typical circulation file provides a record of materials charged out, temporarily in a processing department, or special stack location. This file may provide for some queuing of users to a specific biblio- graphic volume but does not automati- cally balance the waiting users among widely scattered copies which may be duplicated in the total library collection. In addition, such a batch mode circula- tion system does not serve as a locator tool for volumes on the shelf in a geo- graphically decentralized library envi- ronment. From the above example, we can see the need for integration of circulation operations with the total processing op- eration. To carry our example .a step further, the following benefits would ac- 114 1 College & Research Libraries • March 1969 crue from integrating the circulation file function: 1. determining specific location of vol- ume; 2. determining availability by location; 3. queuing of user to item having i~­ mediate availability or shortest wait- ing list or time until return ; 4. triggering order of additional copies according to user demand; 5. handling user cancellations by auto- matically shifting user queues ; 6. determining storage transfers or with- drawals of volumes by their use his- tory according to library standards; . 7. full bibliographic information easily available for the issuance of recalls and overdues. This means that the on-line library sys- tem must have at its heart a file struc- tured to handle both bibliographic and operational data, meeting generally the following requirements: 1. bibliographic information for each volume added to the collection com- patible with national systems require- ments; - 2. subject searching and specific entry searching capabilities; 3. library operational data, including or- dering, accounting, routing, location, processing history, and management information relative to file use and transactions involving use or change of the files; 4. service file information, such as: a. vendor name and address files, b. borrower name and address files (probably the above two are in- stitution-wide shared files) , c. account or fund files, d. user interest profiles; 5. special depth indexed bibliographic files for special subsystems such as selective dissemination in delimited subject areas. As noted above, portions of this file may operate in a file-sharing m~de so that a file connected to one central processor may be accessed for inquiry by another computer over a communication ~ine. This can give important cost savings within an institution. Systems design will evolve which fits individual library needs yet maintains compatibility with national bibliograph~c sources. Since libraries differ vastly m their accounting, processing, and subject matter, these factors , together with size and user services, will determine the system requirements. Therefore, prime systems design factors will be the str~c­ turing of the integrated data base, using as many practical economics, such as file sharing, as possible, yet carrying out required applications and allowing flex- ibility for furth er system evolution and size growth. OPERATING SYSTEM SOFTWARE The operating systems supplied with medium- to large-scale computers most frequently used in on-line commercial systems, enable the computer to receive, process, and transmit data to and from various communications devices con- nected to the system. By buffering and memory paging such software permits the computer to maximize its efficiency, handling many jobs, seemingly simul- taneously. By the establishment of pri- orities in processing and queues of jobs waiting to be processed and then polling these queues , many users of the system can be served with minimum respons e times. It is the development of this op- erating system software together with large mass storage files which has made on-line library systems as possible to- day as the airline reservation systems of this type currently operating and be- ing developed. In fact, in certain situa- tions, on-line processing becomes more feasible from a service and cost view- point than any other type. This ~naly~t believes that in years to come this will also prove true for large research li- braries. Decentralized Research Library System/ 115 Present developments on a national scale certainly point to the time when large libraries will become stations in a national library network system, each re- sponsible for a segment of the world's published literature, by subject or by type. The heart of such a system would depend upon rapid communication be- tween its stations to answer user re- quests. Messages consisting of biblio- graphic information, availability, and facsimile transmission of the document itself will be initiated through use of consoles operating .as part of an on-line system. If these networks are really go- ing to develop, machine readable data bases must be at their heart to enable remote communication and exchange of information. Therefore, it behooves large libraries to develop such systems to pre- pare for national interaction. As oper- ating system software improves, a li- brary's systems design task is strength- ened so that future developments will undoubtedly offer the library a frame- work for even more effective systems. Operating system software is com- plex, extremely expensive to develop, and up to now has been the prime re- sponsibility of the computer manufac- turer. Most installations have added their own improvements faster than they could have become available for release from the manufacturers. Libraries, as an applications user group, will be no ex- ception to this when they operate their own computer systems or share a super- scale system within their parent institu- tions . Libraries will find it necessary to support further operating systems de- velopment and become vocal on their requirements to the computer manufac- turers. Because of the technical lag be- tween what present hardware and soft- ware can accomplish, it may be left up to the library to undertake modifications in operating system software where needed and not wait for the manufac- turer if long systems delays are to be avoided. APPLICATIONS SoFTWARE Programs designed to process infor- mation and produce required output for library staff and users will be the prime responsibility of the library systems group. Certain types of applications pro- grams, SORT /MERGE, for example, may be written for use in the library community by cooperating libraries to achieve specially tailored printed out- put such as book catalogs. Filing stand- ards presently used should be evaluated according to ease of use by the user of the catalog. When more libraries begin to depend upon computer based sys- tems, filing standards deemed .absolute- ly necessary are bound to result. Cer- tainly, it is possible to program the ALA/ LC filing rules, but there is a high degree of variation in present practices in most libraries. A national standard specially suited to machine processing will undoubtedly result in time through the cooperation of the three national li- braries. Some, however, would feel that each library should form its own con- clusions on filing and build from this point initially. In on-line processing systems operat- ing in a time sharing mode, a limited block of core storage will be available for each program during its execution time. Therefore, one application pro- gram characteristic will be structural modularity. Segmentation of programs, small routines, overlaying, and use of common subroutines for programs op- erating in on-line applications will key- note program design. Operating time efficiency .and minimization of core stor- age required will assume increased im- portance. For this reason many programs may be written using assemblers rather than higher order languages. In libraries, as in the computer indus- try, attempts at standardization have not really resulted in much standardiza- tion. Only recently with the formation of EBCDIC codes in use on IBM Sys- 116 I College & Research Libraries • March 1969 tern 360 equipment has the computer industry attempted to offer more than the usual 64 character set. For library use it would seem that standard codes allowing 128 characters would be mini- mal for representation of European al- phabet languages, lower case, and of enough special symbols. With an in- creased use of extended character set chain printers beyond libraries, the manufacturers should be urged to base future hardware designs on eight level codes rather than the various six bit codes now used by all except IBM. It may be that libraries will want to base their systems plans and requirements on the hexadecimal code structure to en- able direct encoding of up to 256 sep- arate characters. Character set develop- ments put into practice by MARC will be watched with interest by all libraries attempting to decide upon basic en- coding techniques or what hardware to buy. Certainly libraries will be faced with generating software to convert ma- chine readable data to fit new expand- ed code structures as new hardware is procured. Software for applications must be the responsibility of the library systems group. In a large system contractors may be employed to write certain soft- ware as specified by the library. Strict control must be maintained, however, to assure a successful end product and the library's ability to take over main- tenance of such programs once installed. This will be one of the most challenging administrative areas for project manage- ment in development of the library's on- line system. GEOGRAPIDC IMPLICATIONS Both national and local geographic implications result from the installation of on-line computer-based systems in large research libraries. National impli- cations, of course, depend upon the role the library chooses in its relation to a national information network. The ap- plication of computers on a batch mode to libraries has brought about the pos- sibility of increased cooperation beyond merely interlibrary loan services. Swap- ping of basic bibliographic information and access to greater index facilities is now possible by construction of suitable on-line communications systems permit- ting remote queiies to search another library's catalogs. Therefore, on a na- tional basis, large research libraries will send and receive queries to search their data base. Thus, some information needs will be met without intervention of more than one librarian in the originating li- brary. This can mean an immense sav- ing in man-hours and effective provision of additional reference librarians with- out the addition of extra staff. On a local level implications of on- line systems techniques will completely change the present trend of library sci- ence toward centralization of processing activities. With centralized processing and subsequent dissemination of the physical volumes to scattered physical locations, cumbersome order and in- process files are created, as well as man- ual material sorting operations, and rout- ing of rna terials to many people, to name only a few salient characteristics. Computer-based batch processing has encouraged this centralization trend to economically justify computer use, maintain audit control of the work proc- essed, and build suitable clerical sup- port necessary for a computer-based system. The advent of on-line tech- niques has made it possible to remove all the delay and inefficiencies of either a manual or centralized batch process- ing mode system. Because of the ability to query a central data bank, with file security provisions built into the system, along with memory protection, it be- comes more desirable to structure large library systems for decentralized acqui- sitions and processing. This would elim- Decentralized Research Library System I 117 inate the physical routing of volumes, cut the cataloging time, and yet give the advantages of centralization with- out the disadvantages. Branch libraries for geographic or subject areas could be built to be serviced more fully than one- man departmental libraries could ever be serviced, by providing longer open hours, more professional help to the user, and more timely delivery of new materials. , In such a decentralized system, each individual branch would have equal ac- cess to query the central data banl<. It would be able, however, to change or add to only its designated portion of the file. Through security .and access codes only appropriate personnel could query levels of management information. File accuracy would be the responsibility of the individual branch library working together with an audit section of the li- brary's systems group, each library hav- ing one representative responsible for solving errors or file problems when they occur. This .audit pool technique as the human interface between error checking software and the individual user of the system has proved effective in many large corporations operating time-sharing systems. In this environ- ment many different departments main- tain such liaison personnel to take care of problems. To summarize, on-line query would give the geographically decentralized li- brary system the advantages of central- ization because of the central data hank of information. It would effectively com- bine the advantages of smaller libraries and decentralized physical processing activities. The disadvantage of cumber- someness on the part of centralization of manual activities would not exist. Secondly, the disadvantage of split rec- ords, ineffective procedural control, and lack of the right operational information in the right physical location would cease to be a problem. For better serv- ice and staff utilization such an on-line system would aid the library adminis- trator effectively to extend and increase service. uSER SERVICES One of the most perplexing problems for the user of .a large research library is the complexity of finding information rapidly and receiving current informa- tion only after delays of six months to a year from its publication. Some reference requests may involve much physical travel between various departments of the library system, even just to find out whether a certain book is in .a particular department. Finding locations of the actual material, once promising biblio- graphic references have been located, consumes much valuable time for user and library staff. In the typical large re- search library reference help is limited and the user is forced to fend for him- self, much of the time inadequately. Li- brarians wish to alleviate this situation, but because of limited staff, certain .an- tiquated procedures, and lack of the proper files in the right places, progress in solving this problem is slow. Use of modem indexing services and techniques, a redefinition of reference service goals, and the design of thor- oughly modern on-line systems designed to free librarians for the most important task in the library-the provision of ade- quate reference service-will alleviate the situation described above. By using such centrally procured services as MARC and MEDLARS, the library will capture basic bibliographic information for a large percentage of references at the time of their order. Upon receipt of an order this information will be added to for more rapid processing of new items. Other factors, such as publisher~s standards, title page standards, univer- sal book numbers, and further delinea- tion of various record types to handle publication types not covered by 118 I College & Research Libraries • March 1969 MARC, will undoubtedly be imple- mented in the coming years. LIBRARY STAFF EFFECI'S Staff members in most large research libraries are familiar with what com- puter systems may do for the library, either through direct experience or through related professional activities. Every library should have a program of in-service education in this area to lay groundwork for systems work being done in the library without arousing ap- prehensions among staff. With this point in mind, on-line systems should have a positive effect on all staff members if the system's capabilities are understood. Library staff must be made to feel a part of any systems development which takes place in their library since they place the requirements in the hands of the system designers. It is said that many librarians are utilized in libraries for work more prop- erly performed by senior clerical per- sonnel. If it be true, their retraining may become a necessity and their em- ployment as librarians in future years called in doubt unless they .are utilized to provide reference, indexing, acquisi- tions, or in managerial functions. Much less professional staff and more clerical staff within the library itseH will need to be concerned with purely processing work. This will permit more librarians to be engaged in the much more crea- tive and professional task of helping the library user to solve his information problems in depth. Librarians should feel much more in control of their libraries, because of the availability of basic information through- out the system, ease of handling many routine requests, time to manage effec- tively, to decide acquisitions, and to make full use of centrally procured pre- processing information. Librarians may turn their fullest attention to the de- sign and provision of new services such as in-depth reference searches, selective dissemination, or other custom tailored information routing. Through effective systems groups trained to develop and maintain the library's computer-based systems, a high confidence level should be achieved among the whole library staff. CONCLUSION As always, librarians are striving to provide increased service to their clien- tele who increase in number .as well as sophistication in their service expecta- tions. On-line systems will do much to make library use easier for the patron and less time-consuming on both his part and the library's; it will also pro- vide a base for the development of na- tional library networks. When each large research library has basic bibliographic data machine-readable and serviced by on-line systems permitting remote query between geographically separated com- puter systems, quicker information proc- essing will result for the information user. A depth of resources will be made available to every library on a faster basis than through the present complex procedures of writing or teletype com- munication and subsequent checking operations performed on each request no matter how routine in nature. Librarians will be free to decentralize ·their collections yet retain the efficien- cies of centralized processing and rec- ord systems. Rapid copy service, fac- simile transmission, microfiche dis semi- nation, and many other techniques will be employed to expand library services to an ever more demanding user group. The system work to now and that planned for the immediate future should effectively prepare the large research li- brary for the development of on-line computer-based record systems for in- formation processing and dissemination. Systems librarians face a great challenge to put together well designed systems and effectively communicate the li- brary's needs to the library's computer center personnel. • •