84VINE

VINE: Volume 14 Issue 1 1984 pp38-42

Page 38

DUTCH AND BELGIAN LIBRARY SYSTEMS: A COMPENDIUM

[Derek Law, Library Automation Controller at the University of Edinburgh, visited 5 operational library systems and one under development during a 2-week study tour of Holland and Belgium. His particular interest lay with the selection, management and operation of library systems - notably for cataloguing - and with the availability and ease of use of online public access catalogues (pacs). The following is a summary account of the 6 sites visited. They cover two quasi-national systems, each with different degrees of government support, and four local systems, one a customised institution specific system and the others turnkey systems. Though the original report deals with them in chronological order of the sites visited, for the purposes of this summary I have re-ordered them to look firstly at the concept of "national" cataloguing networks and then at local systems and how these might (or might not) fit into the national scene. This seems an appropriate treatment in the light of the Australian articles in this issue of VINE. The other main theme I have selected has been the pac and its use. Since this account is already twice-removed from the suppliers of the original information, I would like to apologise in advance to them for any inaccuracies or misrepresentations. These can undoubtedly be attributed to a lack of comprehension on my part. Ed].

Pica Library Network

Pica is the Dutch national shared cataloguing network run by a staff of 30-40 from the Royal Library in the Hague. The system is now in its second incarnation and its next demise (if that is the word) is expected around 1985 when the hardware is scheduled for replacement. Currently, it uses a pair of PDP ll/70s, one of which runs the system, the other of which is used for back-up and development work. The network has 180 terminals spread over 30 locations connected by leased lines. Members include the original eight, i.e. the Royal Library and seven university libraries, plus the Dutch National Bibliography and the Dutch Documentation Centre. This last produces catalogue cards for 900 public libraries. The original concept was of a totally integrated system and a number of applications modules is supported though at the heart is the shared cataloguing system. The database now holds 2.35 million titles, with 2 million shelf-marks, plus 800,000 LC and UK MARC records without locations attached. There is a generous maximum record length of 2,000 characters and, in addition, each user has its own centrally-held file for local data. There is also a separate personal name authority file which, unlike in the WLN system, is not linked to the bibliographic records.

For data entry (i.e. for cataloguing) the system is a mix of the menu-driven (especially for initial screens) and the command-driven. As regards standards, there are no defined input levels but there is a requirement for libraries inputting highest-level records to use the established authority file data. The access paths provided to catalogue records are comparatively restricted and unsophisticated, and are based on the assumption that the enquirer will be working from book-in-hand. Acronym-key (Pica-defined)

Page 39

searching rather than keyword or Boolean is the order of the day. Output is provided in the form of cards, fiche or printed-copy in the case of the Dutch National Bibliography; an impressive feature of the system was the speed with which bulky printed outputs were generated.

Pica also supports:

An acquisitions module, available as a local system or as a shared central service.

A circulation system, using a brief title/author record of 40/10 characters. This was developed in-house to provide a system better suited to the closed-access operating environment of the libraries than the commercially available turnkey packages. Recent modifications have been towards the development of a standardised module, customised through parameter-driving, rather than the original site-specific versions.

A union catalogue sub-system which also supports an inter-lending module. For this, enhancements to the search facilities have been developed to provide better IR access with better search keys and Boolean AND and OR operators. There are, however, political problems here of conflict between Pica and four existing manual national union catalogues, with the result that it is the role of Pica as a shared cataloguing system which is emphasised and promoted. A public access catalogue, based on Boolean logic, is under development and is planned for introduction in May 1984.

The charging structure appears to create problems for heavy users. The capital cost of the central configuration was funded by the Ministry of Education, but Pica must recover all costs of holding the records on the database and running the system; libraries, therefore, pay the actual charge for all their records held which has, for example, created very high overheads for a library such the University of Utrecht (see below for its approach).

NEWWAVE at the Royal Library/ Brussels

This is not strictly a national cataloguing network; rather it is the internal system of the national reference library, the Royal Library, which has to date unexploited potential as a national shared network a la Pica. Unlike the Dutch situation, the political will to support such a commitment appears to be lacking. The original concept of NEWWAVE in the late 70s was for an interactive real time cataloguing system with powerful retrieval facilities.

The system runs on a Siemens 7531, under the library's own database management system though a switch to SESAM was planned for the first quarter of 1984. There are 30 terminals connected, all located within the Royal Library. All software was written in-house, in assembler, and it is managed by the Centre National de Documentation Scientifique et Technique. The records are based on ISBD with an associated authority file, and stored in a UNIMARC-compatible format. To date, the database comprises 400,000 bibliographic descriptions and 200,000 authority file records for all data, whether publishers, personal or corporate authors, subjects, uniform titles. Plans are in hand to turn the subject authority records (LCSH, translated into Dutch and French and also the

Page 40

original English) into a separate file. Retrieval capabilities are powerful, allowing searching (and browsing) in the indexes, via these in the authority file and the bibliographic descriptions, with two-way pointers or linking between all files allowing users great flexibility in moving through the system. There is also Boolean AND and OR searching and truncation for stem searching.

For output, the confinement of the network to the Royal Library has resulted in concentration on a wide range of hard-copy offline products, from SDI listings to COM and the production of the photo typeset Belgian Bibliography. There is no pac and the system as it is is unsuitable for public use; however, some terminals are now being put in public areas but with trained staff to act as intermediaries. No other modules are supported and any such developments, e.g. for circulation control, have had low priority since the Royal Library itself has had little requirement for them.

Though not primarily designed or presently intended as a national shared cataloguing network, the system does have potential as such and recent enhancements have been for the provision of local data so that NEWWAVE could be run for a group of libraries.

University of Utrecht

Utrecht is a Pica user and, as many readers will be aware, a Geac library. It provides a study in the potentially conflicting demands of belonging to a shared "national" system and meeting local needs via a local system. Utrecht was reorganised in 1975 from 14 libraries into one large one. Since then, it has built up on file in the University Computing Centre machine-readable records for some 500,000 monographs and 140,000 serial titles, covering nearly 2 million volumes. This file was used to produce catalogue products for the multiple sites of the library. Now, however, it is becoming too cumbersome for the Computing Centre and all records will be loaded onto the Geac and accessed via public query. Utrecht's Geac network has 90 terminal connections and a second processor is in prospect. It is a closed access library so catalogue consultation is a necessary prelude to borrowing; public query, therefore, features large in Utrecht's priorities. The system is menu driven and users seem to have adapted well to it, though Utrecht has inserted additional HELP screens, as an aid to defining types of search. The Central Library has 9 public query terminals out to public use. For cataloguing, the intention is to catalogue on the Geac; however, in compliance with government policy, Utrecht has agreed to put 10,000 records through Pica in 1984, at a cost of 5 guilders per record. These records will concentrate on areas where the Pica database is weak. For Utrecht, the main benefits of Pica lie not in its use as a shared cataloguing resource but in its interlending potential, a more politically sensitive issue.

SAILS

S.A.I.L.S, an acronym hitherto unfamiliar to British libraries, is the second of the turnkey systems. SAILS is the Swets Automated Independent Library System, developed from the automated in-house system of the Swets and Zeitlinger subscription agency. It is to be sold as a totally integrated

Page 41

package. Though developed and run on an IBM at Swets, the emphasis is on complete portability and the system is currently undergoing field trials on a VAX 730 at an Irish University. These trials will be extended to one or two other European libraries of different types before the system is marketed, hopefully at the end of 1984. The system is menu-driven with the emphasis very much on user-friendliness and user-adaptability. Thus screen layouts can readily be altered to suit individual requirements; menus are used extensively to guide choices, with help screens for each menu option; though MARC-compatible and available for display with MARC tags attached, records are normally created using natural language prompts. For access the system supports both Boolean and keyword searching. All library functions can apparently be provided viz. acquisitions, accounts, cataloguing, circulation, ILL, serials control, exchanges and public enquiry though this last is not yet in a settled form. The system design is of a simple integrated database with all applications/ functions addressing the central bibliographic record; once this has been identified as a result of a search (by whatever access point) the user can move around the database retrieving relevant data from applications module to module.

VUBIS

VUBIS 1, now in the sixth year of its life, was developed at and for the library of the Vrije Universitat Brussels (Free University of Brussels). Staff are now looking towards VUBIS 2 which will be an extended system for shared cataloguing and pacs involving the Technische Hochschule and the Public Library System in Eindhoven, building on the best features of VUBIS 1. Interestingly, the opac facility was an integral part of VUBIS from its inception and has not developed or evolved as a by-product of other functions. The VUBIS 1 runs on a Dec PDP 11/60 with 35 Kb of memory under the MUMPS operating system. A major factor in the change to VUBIS 2 is the need to switch to a modern, standard version of MUMPS which will make future support and upgrades easier. The present system can support from 2-25 terminals and the University currently has 4 public-access terminals in the Main Library and others on two more campuses. Catalogue records are created online, in-house using a local record format with fields prompted for in natural language. Title is entered first and the cataloguer may bridge to search the database for presence of duplicates. The author field allows a maximum of 3 authors at present, but this will be extended for VUBIS 2. Subject data is a mix of UDC plus a controlled tri-lingual thesaurus of UDC terms. Catalogue records are sent to a common work file and this is merged overnight with the master-file every two or three days. Annual throughput for the 4 cataloguers is 10,000 titles.

Public access is the most important feature of the system and has already been fully described in the literature. (1) In brief, searches may be conducted in one of three languages, Dutch, French and English. The system is menu-driven with automatic time-outs if no keys are depressed within a specified interval. Most access relies on a two-tier approach, useful for retrieving inexact or inaccurately remembered items but irritating for exact author or known-item searching. For example, an author search displays first a list of full surnames matching the search term; for the surname selected, forenames are

Page 42

then displayed with postings. Once the user identifies the correct author, the system moves on the individual bibliographic records. There is no facility for skipping this and going directly to a known author or to a known record, though VUBIS 2 will make this available. Subject searching is via an index of keywords attached to broad categories of UDC numbers. The major disadvantage to this is that, despite good intentions, the three language thesaurus does not contain equivalent terms in each language, nor are these equivalents, where present, linked.

DOBIS/LIBIS

The final system was the IBM DOBIS/LIBIS system at the Katholiek Universitet Leuven (KUL). KUL was formally founded in 1970 after a split, in 1968, with the original, 15th century Leuven University. Even so, KUL is a large organisation, with five major processing units and eighty departmental libraries. DOBIS/LIBIS is a joint development with Dortmund University; IBM acquired the distribution rights and to-date sixty licenses have been sold world-wide.

The system runs on an IBM 3033, using about 15% of that computer's capacity, with IBM 3270 terminals. KUL's configuration is for 70 dedicated terminals with 17 leased lines to branches and a port for dial-up access. The 990 MB of disk store hold 500,000 records online. The mainframe supports in addition 150 other terminals all of which may access the library's files to an agreed level.

The system has grown up from a base of cataloguing, though other applications are emerging to create an integrated system. Acquisitions is undergoing final trials, periodicals control was released in 1983 and is at the field-trial stage, and there are plans for an archives management system, electronic mail, and book reservations for closed access. The normal search features of the system are very powerful with Boolean and stem searching, and modification by fixed-field data e.g. document-type, date etc. On displayed listings of matches, the nearest hit is always displayed as second on the list; when displayed first, users tended to page back to discover the previous entry.

Title searches are fundamentally KWIC searches with postings for terms given prior to combining them. On subject search, the system allows very large numbers of records to be retrieved by writing them to a work-file which can then be re-searched under a modification of the original search term.

Catalogue entry may be menu or command-driven, depending on the skill of the user. All items are catalogued and entered to the database locally at a current rate of 90,000 items a year. Although online public access has been available since 1981 (there are 4 pac terminals in the Main Library), there is still a fair degree of batch output. The system can produce cards, printed copy, such as Accessions Lists, and COM.

Reference

(1) Gerrit Alewaeters et al VUBIS: A User-friendly online system. Information Technology and Libraries, Sept. 1982 pp.206-221.