Library Innovation: Experiences at Tilburg University

Hans Geleijnse, Hans Roes

Tilburg University Library

HTML version of an article published in the Encyclopedia of Library and Information Science, vol. 58, supplement 21, pp. 113 - 134, 1996.

I. Introduction
II. Initial position
II.A. Tilburg University and its library in the 1980's
II.B. A new building and an innovative program
II.C. Fundraising and partnerships
II.D. Managing the innovative process
III. Projects started under the HT DIC program
III.A. Integrated Desktop
III.B. KUBguide
III.C. Lendomat: self-service circulation
III.D. Online Contents
III.E. European cooperation: Telephassa conferences
IV. Current projects
IV.A. Ariadne: towards electronic document delivery
IV.B. KWIK: porting the Mercury interface
IV.C. ELISE: coloured image databases
IV.D. Grey Files: research papers online
IV.E. ERA: Electronic Reference Aid
V. Conclusion and prospects
V.A. Evaluation and success factors
V.B Future prospects
Bibliography
Author Information
Acknowledgements

I. Introduction

This article describes the experiences of a university library in the Netherlands with innovation. In 1989 Tilburg University launched a comprehensive plan for a new library building and the development of new library services, the High Tech Documentation Information and Communication program (HT DIC). Section II covers the initial position. It describes the university and its library as it was just before the program started. Also the central themes and principles of the program as well as managerial and funding issues are discussed here.

Section III describes the major projects which started under the HT DIC program: desktop integration and knowledge navigation; a self-service circulation tool; a current awareness service and European cooperation in the Thelephassa conferences. Section IV goes on to report on current projects in the area of electronic document delivery, an integrated user interface for heterogenous information systems; coloured image databases; electronic availability of preprints and the development of a tool for electronic reference aid.

Section V concludes with an identification of critical factors throughout the innovative program. The article finishes with a short look into the possible role of libraries in the uncertain future.

II. Initial position

II.A. Tilburg University and its library in the 1980's

In 1989 Tilburg University, The Netherlands, launched a plan in which the building of a new high-tech library and the development of innovative information services was announced.

At that moment the University Library had six centrally managed departmental branches on different locations on the campus. It was a medium-sized library with approximately 700,000 volumes and 54 full time equivalent staff. The first wave of library automation was completed. The university participated in and contributed to the Dutch PICA shared cataloguing system and national Inter Library Loan system. The PICA Local Library system was implemented for acquisition and circulation, campus-wide access was provided to the local OPAC.

Beside that, the University Library staff had already some experience with in-house development of databases. Since 1985 special databases with respect to particular parts of the collection had been developed, hosted and maintained. An important example is the Excerpta Informatica Online Database with abstracts produced by library staff. This database could be developed into a national focal point with respect to applied computer science mainly because of the expertise of the staff and the quality service that could be provided.

The library is part of a university, founded in 1927, with special emphasis on social sciences and the humanities and with a limited number of faculties: economics, social sciences, law, linguistics and philosophy. Currently 10,000 students are enrolled, 1,600 staff are employed. Since the early stage of library automation, the library has a close cooperation with the university's computer centre.

In 1989 an audacious plan was published aiming at the development of a future oriented, highly innovative library that would be an asset for the university, an attraction for students and a excellent support facility for teaching and research. The idea to give the library a central role on campus fitted completely in the strategic plan of the university. A key element in the strategy was a strong emphasis on quality in education and research: every single department should be ranked in the national top three in their particular subject area. A concept of student-focused learning was announced. Special emphasis was given to the use of information technology in order to improve the infrastructure, the supporting facilities and the learning and research processes.

II.B. A new building and an innovative program

Immediately after the decision of the Dutch government to grant subsidy for a new library building at Tilburg university, the library started to make a comprehensive program for this building. In detail the requirements were described not only with respect to stacks and study places, but also with respect to the computer infrastructure and the consequences for heating, cooling, ducts, electricity, sunrays, ozone etc. Important principles in the program were:

Flexibility: the three floors of the library should have a minimum of fixed elements. It should be possible to realize important changes in the future within the building: for instance to have more study places and fewer volumes in open access will be a realistic option for the future.
User Orientation: the library should have a pleasant and inviting atmosphere where users can browse through the bookstacks but can also do their work, singly or with others, using books and/or computers.
Network Connectivity: every single desk and every single study place should have ducts and should possibly be equipped with a personal computer or a UNIX work station. This means that at any moment a computer can be installed and directly connected to the university local area network.
Air-conditioning: because of the presence of several hundreds of computers and other equipment, like printers, scanners, xerox machines, but also because of the requirements for preservation of the collection, special demands with respect to the climate in the building were necessary. Also special sun protection was required.

In general these principles were maintained and realized. The work on the new building started summer 1990 and was completed in January 1992. At the moment the whole collection of 700,000 volumes is housed in one building, with 200,000 books in open access. The building provides facilities for about 80 staff members and offers 900 study places.

The ideas with respect to the building were very much connected with ideas on the role of an academic library in the future. In close cooperation with the university's computer centre a blueprint was made for new library services and for new ways of information handling and information production. Based on this blueprint an innovation program was launched with the wholehearted support of the Board of Governors of the university and in close cooperation with Digital Equipment Corporation as the most important industrial partner in the program. Key elements in this program were:

To make full use of information technology in order to improve library processes and library services.
Electronic information will be of growing importance. The library - in conjunction with the computer centre - should provide facilities to handle this information. It should be irrelevant whether the information concerns text, data, voice, still or moving images, where and how the information is stored, local or remote. Users should have access to (multimedia) information from their desktop at the university and at home.
The consumer of information in a university environment is often also the producer of new information. Researchers and students conduct database searches, retrieve information, use this information, add results of new research and new ideas to it in order to produce new information, new articles, new theses and papers that can be discussed and communicated with colleagues and others. The library should support the productivity of the university and should facilitate this process. An important way to do that is:
Desktop integration. The online library information should be integrated with other computing services like database management, electronic communication, word processing, statistical and graphical software, and printing in order to support education and research in the university.
The technical infrastructure should be characterized by flexibility and stability, and should be based on technological standards and open solutions.

With this concept several project teams with staff from the library and the computer centre, sometimes assisted by specialists from other departments, started their work in September 1989 and completed their first results in Spring 1992. The results were achieved mainly by the work of internal staff of the library and the computer centre, taking advantage of a well established cooperation between the two units and the experience of various staff members of the library as, starting in 1985, the library had been developing its own applications in specific areas and had already prepared its staff through extensive educational programs. The creativity and expertise of staff can been regarded as the basis for success.

II.C. Fundraising and partnerships

From the beginning the program to build a future oriented library was strongly supported by the Board of Governors of the university. It fitted completely in the strategic plan of the university and should stimulate a new way of learning and working, making full advantage of a well elaborated technical infrastructure and campus wide access to online information.

The university was convinced that an innovative program with an important emphasis on the implementation of new information technologies could only be realized with the help of other parties. Because of that, there was a strong and clear preference for cooperation both within the university and with other parties: libraries, vendors, publishers and others.

One of the first steps was to look for a strong commitment of an important hardware supplier. It turned out that Digital Equipment Corporation shared the vision of the university and was prepared and able to form a close partnership with Tilburg University and to invest in the realization of the program. Digital supplied a large amount of technological expertise as well as financial means for making the program a success.

Because there was a stimulating and futuristic vision and a well established partnership the Dutch Department of Education and Science was willing to sponsor not only the costs of a new building but also to cover a major part of the initial costs of several new projects. This sponsorship was continued after the completion of first results which showed the government that the university was capable to carry out the projects according to schedule.

An agreement was made with PICA, the Dutch organization for library automation, which stated that PICA would provide financial support and experts who would take an active part in some projects. Cooperation also took place with major publishers such as Elsevier Science and Wolters Kluwer Academic Publishers.

It was never anyone's intention to execute these pilot projects in isolation and therefore cooperation was established with other university libraries. A European project in the framework of COMETT (The Telephassa project) was realized with the Universit t Autonom de Barcelona (Spain) and the University of Patras (Greece). A memorandum of understanding was signed with Carnegie Mellon University (Pitts- burgh, USA) in order to exchange ideas and information. A pilot project on Online Contents was developed in cooperation with the Dutch Royal Library in The Hague and sponsored by the Dutch Department of Economic Affairs.

Overall, the initial out of pocket costs of the program could be covered by external funding. The university contributed heavily with extensive personnel efforts through the participation of staff of the library and the computer centre. The total budget for the library building was Dfl. 25 million. The overall costs of the improvement of the technical infrastructure of the university, the computing facilities for the library, the implementation of 1,500 integrated desktop computers on the campus and the costs of various innovative projects in the framework of the program for the new library were approximately Dfl. 20 million. Included are the costs for hardware and software and costs of staff.

Beside that, Tilburg University had to guarantee that all new equipment, new software and new applications in the future could be maintained and replaced in due time. For that purpose the University Council decided - already in 1989, just before the first project started - to allocate structural funds and to build reserve funds for this purpose.

II.D. Managing the innovative process

In Autumn 1989 seven project-teams were installed. The first task of these teams was to examine possible solutions for seven key issues which were identified in the program published in Spring 1989. These issues were scanning and optical character recognition, office automation, databases, networking, learning environments, printing and self-service circulation.In these teams, staff members of the library and the computer centre, faculty staff and consultants from Digital Equipment worked together. A Program Management - the management of the library and the computer centre and the project managers - evaluated the results of the seven project teams. The management concluded that there was full agreement on the vision as it had been made explicit in the program, but recognized two important problems.

First of all, the actual budget was limited. It was obvious that not all of the possible projects could be launched immediately. Secondly, the environment - university staff, personnel of the library and the computer centre, students - needed a clear perspective of future developments and of the way information provision within the university would evolve. They had to be convinced in a concrete form, that bright ideas could become reality. For these reasons, the Program Management decided:

To develop a demonstration model in order to give users and sponsors an impression of the impact of the program. For that purpose dedicated equipment was installed in a special room at the university. Presentations and demonstrations for university staff and for external visitors and sponsors could be organized there to visualize the idea of integration of data, texts, and images.
To focus on the realization of clearly defined services like an Online Contents database, a Campus Wide Information System, the design of the Integrated Desktop, networked printing facilities and a self-service circulation system.

For each of these topics and for the preconditions to realize these services - like a proper database management system, network navigation tools and the physical network and other infrastructural facilities - new project teams were installed in order to develop the deliverables as they were specified in an action plan. Each project team would have one or more working groups in a specific field. In these teams and groups, library staff worked together again with staff from the computer centre. In some cases, staff from other university departments joined the teams as well.

The projects were realized without the use of any official standard method for system development. Functional and technical specifications were set. Subsequently, an experimental prototype was built. This prototype was tested and validated. Important principles were:

To buy, whenever possible, what is available on the market.
To look for open solutions preferably based on (international) standards.
Cooperation and communication between members of the project teams through electronic conferencing.

In April 1991 a large scale demonstration of the current status of the various projects could be organized. In May 1992 the new library with the first innovative services in operation was opened by the Dutch Minister of Education and Science.

III. Projects started under the HT DIC program

III.A. Integrated Desktop

Desktop integration was and remains one of the central concepts in the Tilburg innovation program. The aim is a user desktop from which information can be retrieved, processed and edited to create new products of scientific information, a concept also known as the scholars workstation. The Tilburg implementation of this workstation is a networked DOS personal computer with an MS-Windows graphical user interface, which is still evolving towards tighter integration of functionality. Basic integration is achieved via the MS Windows environment. For processing and editing each student and scholar can use software for which campus-licenses have been arranged. These include Wordperfect for word processing; an electronic multilingual dictionary; SAS and SPSS for statistical processing; the Quattro Pro spreadsheet; linear programming software; dBase and Paradox for database applications; Harvard Graphics for presentations; network software including Telnet, FTP, electronic mail and a newsreader as well as software for consulting networked CD-ROMs and the KUBguide for access to local and remote databases (see next paragraph); a host of tools varying from anti-virus software to decompressing, encoding, decoding software and tools for easy formatting of diskettes; and finally a set of computer based instruction programs to explain some of the features of the integrated desktop and the services which can be reached via this desktop. Most applications are available in both DOS and MS-Windows versions, the tendency is though for the latter to gradually replace the former. Printing is a departmental affair for staff while for students there are three decentralized printing facilities in the library which are mainly operated by self-service. To this end, each student has a printing account in which he or she can put deposits at the main desk in the library. Print jobs are issued to a queue and can be released using a password on a machine located next to the printer.

Figure 1 The Integrated Desktop

In total there are about 1,400 integrated desktops for staff and faculty and about 700 for students, of which 450 are located in the new library building. These latter 450 have special security provisions preventing the spreading of viruses and limiting the danger of disk errors to a minimum. This is achieved via special boot-PROMs which ensure that:

only Tilburg University students which can identify themselves with a username and password have access,
these users cannot start their own programs, a condition which is not imposed on the desktop of staff which is of course more a `personal' computer,
every time a user logs on he starts with a `clean' desktop, implying that no files of prior users hang on,
every day the hard disk is formatted automatically to prevent viruses from spreading.

Students can make reservations for these integrated desktops via public terminals. During semesters on average 90 percent of the desktops are in use.

A recent survey showed that not all possibilities of these advanced desktops are exploited in full. That is why a project has started to encourage use of the possibilities of this toolbox in the curriculum. In two departments of the university small scale demonstration projects are set up, tailored to the needs of the students in preparing papers and presentations. Library personnel will develop these programs in close cooperation with teaching staff. If necessary, help from the computer centre is available.

An example of the strength of the integrated desktop is the way in which this article evolved. The authors communicated their contributions mainly via electronic mail and exchanged several Wordperfect versions of the paper, while other members of the library staff gave comments on final drafts, again through electronic mail. The bibliography was compiled from local databases and imported into the final document, while illustrations were either re-used from prior publications or newly produced using Harvard Graphics and screen capture programs.

III.B. KUBguide

Knowledge navigation is the second central concept in the innovation program. It means the guidance of users at any level of experience in a world of rapidly growing possibilities for heterogeneous networked information services. Besides the OPAC, Tilburg University library offers its users access to about twenty other information services, varying from a Campus Wide Information System to abstracting and indexing services in the field of computer science (the Excerpta Informatica Online Databases), a database for research papers or preprints, and the Online Contents service to be described below. Internet access is available through Mark Resmer's LIBS, giving access to OPACs all over the world, and through Gopher. World Wibe Web browsers, probably Lynx and Mosaic (directly via the integrated desktop) will be added soon. Library services in the Netherlands can be reached easily through the OPAC via the Dutch Open Library Network. This not only includes access to OPACs from other university and major public libraries in the Netherlands, but also the Dutch union catalogue as well as a national Online Contents database, covering 12,000 journal titles in frequent demand in Dutch Inter Library Loan traffic. Late 1994 a test started with access to OCLC's FirstSearch services (Worldcat, Article1st etc.) relayed via the Dutch Open Library Network.

The first step in developing navigational aid in this still growing heterogenous world has been KUBguide, KUB being the Dutch acronym for Tilburg University (Katholieke Universiteit Brabant). This is a menu- driven, bilingual terminal interface giving short help in selecting relevant services. When a user has made a choice, connection with the database, no matter where it resides, is automatically made. A KUBguide screen is displayed in figure 2. Here the left part of the screen lists the services available, while the right part of the screen contains a short description of the highlighted service.

Figure 2 KUBguide

There are several technical and conceptual drawbacks to the KUBguide approach which hint in the direction of solutions which go from guided network navigation towards knowledge navigation.

First of all KUBguide is a terminal interface, implying that the service leans heavily on central computers. With use growing at about 30 percent per year this would mean investing in ever larger, and more expensive equipment. The solution is client/server technology enabling the use of the power of the integrated desktop more efficiently as well as enabling a cheaper path of growth by introducing extra servers for local databases as the load grows. An additional advantage is of course the possibility of a graphical user interface running on the clients.

A second, inherent but major, drawback is that the databases to which KUBguide connects have heterogenous interfaces. In a sense, KUBguide gives access to a complete archaeology of library information systems with all their different possibilities and interfaces. Also, databases on CD-ROMs are not well integrated, since these are applications in the PC/Novell world and are therefore accessed directly from the integrated desktop and not via KUBguide.

The third drawback is that support for users in selecting relevant databases for their searches is only rudimentary. Either databases should be integrated or clustered, or the level of guidance should be enhanced. The second approach is investigated in Tilburg since this route allows for the inclusion of databases in searches not only covering local databases but also, and perhaps increasingly so, resources on the Internet.

A final drawback is that the integration of heterogenous full text representations (black and white and coloured images, ASCII/SGML, Postscript etc.) cannot be achieved.

These problems are addressed in projects currently under development at Tilburg University and described below: client/server, graphical user interface in the KWIK project, integration of full text in the Ariadne and KWIK projects, selection of resources in the Electronic Reference Aid project.

III.C. Lendomat: self-service circulation

The lendomat project addressed the automation of a basic library function, circulation of books. As of March 1992 Tilburg University Library is the first European library with a completely automated self- service circulation system. With a subsidy from the Dutch Department of Economic Affairs, the Lendomat was designed and realised based on a combination of existing technology through a cooperation of Tilburg University, PICA, a security firm and a software developer. In this sense it was one of the projects with most involvement of external partners.

The lendomat uses several functional components:

Users identify themselves by inserting a chipcard into the lendomat which can also take care of billing if necessary.
Books are identified by scanning barcode labels.
The lendomat combines book and user data and communicates with the local library circulation system, which is in turn connected with the OPAC showing if books are on loan. Users can renew loans and make reservations via the OPAC also.
The lendomat deactivates a security strip if communication with the local library system shows that the user is authorised and the book may be circulated. Users receive a ticket with title and date when the book is due for return.
If the book is returned to the library, a return-lendomat checks in the book automatically, again via communication with the circulation system which also updates the OPAC information, user status, and activates the security strip again. Again, a user receives a ticket, now as evidence that he returned the book.

Users are not obliged to make use of the lendomat, but in practice nearly all books leave the library via the machine which operates very easily.

III.D. Online Contents

The basic rationale for the Online Contents project stems from three observations. First of all, the bulk, approximately 70 percent in Tilburg, of a university library's budget is spent on journals nowadays, reflecting the ever growing importance of this literature in scholarly communication. Secondly, library automation had reached the stage of campus wide accessible OPACs, but OPACs usually do not show the content of journals, they merely show journal holding information. Thirdly, alternatives for searching journal literature are either not timely (abstract journals) or are expensive, hard to budget and often not adequate for end user searching (online abstracting and indexing databases). The third problem has diminished since the proliferation of databases on CD-ROM, although the timeliness problem remains with CD-ROMs. But abstracting and indexing services do not necessarily map with a local journals collection, leading to a possible underutilisation of the library's collection. The Online Contents project sought a solution to these problems in a cheap way.

The main costs in producing bibliographic or reference databases are incurred with data entry. Especially when records are enhanced with keywords and abstracts the personnel costs rise steeply. On the other hand, scholars and skilled library staff are often capable of judging the value of a reference by title of the article, author and journal title. In fact, very often scholars routinely scan certain journals for new leads. Why not put this information on the desktop ?

The Online Contents database at Tilburg University holds now over 300,000 records describing in a bibliographically sparse way articles published in approximately 1,800 journals to which the library subscribes. At the start of the project, in 1990, the tables of contents of all of these journals were scanned upon arrival at the library. Overnight these images were translated into machine readable records using OCR (optical character recognition) software. The next day these records are edited and tagged and ready as database-input. At the time the journal issue hit the library's shelves it was also present in the database. This production load could be handled with approximately 1.5 full time equivalent.

A second way of obtaining cheap data entry was sought in cooperation with publishers since scanning and OCR'ing table of contents pages is putting something into machine readable form where it probably exists in machine readable form with the publisher. This idea was worked out in close cooperation with Elsevier Science who deliver on a monthly base SGML (ISO 8879/Standard Generalized Markup Language) marked up records of articles published in about 60 journals in the collection. These records contain also abstracts and keywords as appearing in the journals.

Figure 3 OLC record / ESP record

The Tilburg Online Contents service triggered a national service in the Netherlands. In cooperation with the Dutch Royal Library and PICA, the specifications for a national union database of journal records were worked out. This led to the national Online Contents database which is being fed with data produced by Swets subscription agents in the Netherlands. Currently, Swets delivers data on 12,000 journals to PICA. Tilburg has stopped producing records of about 1,100 journals which can now be obtained through PICA.

Use of the service in Tilburg is intense and still growing and users report satisfaction with the tool in their daily work. The service will undergo some major changes though in the future. First of all it will be transferred to a new database engine for performance reasons and a new interface (see the section on KWIK below) to offer a uniform interface to the heterogenous existing reference databases. Secondly, the service will be enhanced with document ordering and delivery facilities (see the section on Ariadne below) at first and document viewing and printing in a later stage (see the section on KWIK below). A third enhancement is and SDI (selective dissemination of information) service based on user profiles, with which already was experimented. About 60 faculty were involved for which reference staff developed profiles. These profiles ranged from very simple `baskets' of journals to sophisticated subject oriented queries. Despite the fact that in most cases only article titles are available, users were pretty satisfied with recall rates. The profiles were matched against the daily updates of the database and in case of hits these were captured in a file which was sent to users by electronic mail. Due to the decision to change the search engine, this experiment was interrupted, but will certainly be pursued again in order to be developed to full operational level, since this kind of personalized services is, in our opinion, the core business of libraries in the future.

III.E. European cooperation: Telephassa conferences

In the framework of the COMETT program of the European Community a project was executed by Tilburg University in cooperation with the Universit t Autonom de Barcelona (Spain), the University of Patras (Greece) and Digital. An important goal of this European project - known as the Telephassa project - was the organisation of four conferences on innovative information services and information handling, each of them to be held in Barcelona, Patras and Tilburg.

Key issues in these conferences were the rapid changes in information technology and its effect on libraries and industry. Important aspects were the need for information exchange between libraries and industry, and between libraries from different parts of Europe in order to disseminate information and stimulate new developments with respect to the use of information technology in libraries.

Topics of the seminars were database development, networks and infrastructure, electronic document delivery, knowledge navigation, strategic planning, human and organizational aspects and discussions on the future role of libraries.

The conferences were organized between Autumn 1991 and Spring 1993. In total 452 persons from 144 different organizations in 18 countries attended these conferences. The Telephassa project certainly gave a stimulus to important developments in libraries in various European countries and contributed to the dissemination of the ideas developed at Tilburg University.

IV. Current projects

IV.A. Ariadne: towards electronic document delivery

An important goal in innovating library processes is to bring the library to the desktop of the user. This means that a user should not only be able to search for references in local and remote databases, but she should also have a transparent opportunity to obtain the full text information of the articles (books are not considered here) she found in the process, no matter whether the article is present local or remote. As a first step the information should be easy to order and preferably quickly available in printed form. As a next step, the information should be available electronically at the desktop, where a user can then decide whether or not to make a printout. For the user these processes should be totally and transparently integrated with the search process. Another important factor to consider is that a system for electronic document delivery should enhance the efficiency of library processes. Here storage of electronic copies is important since such copies can be reused.

Based on an analysis of existing systems and current document delivery activities in the library, the following requirements for a system for electronic document delivery, named Ariadne, were drawn up. First of all the system must be self-contained in order to meet internal demand for document delivery in a local area network environment. This implies that a user must be able to invoke the system from (a) local reference database(s) where a set of pointers to articles to be ordered has been constructed.

Secondly, the system to be developed must be able to communicate with equivalent systems in a wide area network environment. This implies that the user should not need to worry whether an article is locally available, the system should be able to forward the request if it can only be obtained via ILL, and, conversely, the system should be able to receive and handle requests from remote document delivery systems.

Thirdly, the system must be standards compliant in order to protect the investment in development and make cooperation with other libraries possible. Notable standards are ISO-ILL covering inter library loan request, GEDI, an extension in development of the ILL standard for electronic document delivery, and ISO-BIBLID, a standard for unique identification for contributions in serial publications.

A fourth requirement is integration of the system with existing library services (reference databases) and (document delivery) procedures.

Fifth, the system design must be prepared for future developments, especially in the realm of document representations. As a start, the system is designed to work with images (electronic copies) produced by scanning from the hardcopy issues of journals, but other full text standards are developing, notably SGML. Another example is electronic journals developing on the Internet in an already wide variety of formats. These new developments should be easy to incorporate.

Sixth, where image production is concerned, the organisation of the scanning part of the system should be preferably based on a demand driven approach, although a supply driven approach should still be possible. Demand and supply driven approaches are not mutually exclusive. Some material might already be available in electronic format, for example in the case where a license agreement with a publisher is concluded, while the system should also be able to react proactively. E.g. if it is known, based on management information to be generated by the system, that articles from certain journals are in frequent demand, a decision can be made to scan these journals in advance.

Seventh, and closely related to the demand driven approach: storage of articles is an option, not a rule. In this respect, images (or more general: electronic representations) of articles will only be stored whenever there are no legal motives not to do so.

Eight, users should be able to demand output in different formats: printed output for traditional (snail) mail delivery; electronic delivery to an end user's workstation to be viewed from screen; and facsimile delivery.

Finally, integration with ILL services and procedures must be accomplished and it must be possible to transmit electronic documents across a wide area network to equivalent systems.

All in all these requirements describe a complex system. Its environment is represented in figure 4.

Figure 4 System Environment Ariadne

Ariadne has been developed over the last 18 months in close cooperation with PICA to a more general system called the Document Delivery Server (DDS). This DDS, in conjunction with a Document Workstation (DWS) developed by RLG and based on the Ariel document delivery station, is the basis for a rapid document delivery system in the Netherlands. In this nation-wide system the DDS integrates the various components: local OPACs, the national Online Contents database, the PICA ILL system and of course the DWS and other DDSs. In university libraries with multiple locations, the DDS also takes care of routing requests to the appropriate departments. In the national system the option of storing articles is ruled out for legal reasons.

The Tilburg DDS, Ariadne, will have an additional task in realising the ultimate goal of a comprehensive and integrated document delivery system on our campus. As a start, the existing cooperation with Elsevier in producing the Online Contents database is enhanced to cover electronic copies of articles published in some 100 Elsevier journals to which the library subscribes. This project, which has been labelled EASE (Elsevier Articles Supplied Electronically), is the first large scale experiment of this kind in Europe and resembles the TULIP project under development in the United States. Images of articles are produced by Elsevier and stored in an image database connected to Ariadne. From October 1994 on, users are able to request printouts of Elsevier articles dated January 1 1994 or later to which they found references in the Online Contents database. When they request these printouts they can designate them to local printer queues. From early 1995 on it will also be possible to view these images on their desktop after implementation of the KWIK-GUI (see next section). This experiment is closely monitored to obtain as much information as possible on user appreciation and to test the economic consequences of licensing on collection development. Collection development will also be influenced by the fact that for the first time hard data on journals use can be obtained.

IV.B. KWIK: porting the Mercury interface

As stated in section III.B, one of the drawbacks of the current KUBguide is that it is mainly a gateway to a world of heterogenous databases with, probably worse, heterogenous interfaces. Principal goals of the KWIK (Dutch for the metal mercury) project are to develop a uniform graphical user interface to a world of possibly heterogenous database engines, and to introduce client/server technology so as to reap the benefits of the power of the installed base of integrated desktops, while at the same time enabling a steady and affordable growth path for database servers.

The principal standards for implementing client/server in information retrieval are ISO Search and Retrieve and ANSI Z39.50 protocols. The latter protocol, although not very different from its ISO counterpart is by far the most popular, and certainly the most discussed at the moment. One of the first implementations of the Z39.50 protocol was realised in Carnegie Mellon's Mercury project with involvement from Digital. The Mercury project developed Z39.50 database servers and X- Windows clients in a UNIX environment. Through Digital a cooperation was set up to port the Mercury interface from X-Windows to MS- Windows clients and to test the integration of different database engines with the Z39.50 glue layer developed at Carnegie Mellon.

An impression of the interface is given in figure 5. It is an easy to operate point and click interface, while the underlying MS-Windows technology enables a much tighter integration than could be achieved with terminal clients in DOS boxes. Selecting databases, formulating queries, viewing records can be done in an intuitive way. Another advantage is the possibility to integrate full text in various formats. As stated above, TIFF images of Elsevier articles can be viewed from the KWIK interface. Extensions will be viewing of JPEG colour images (see the section on ELISE below) and viewing of Postscript files (see the section on Grey Files below). All this can be done in a transparent way by clicking on an image button. The KWIK interface will become operational on an experimental basis from early 1995 on.

Figure 5 The KWIK interface

The KWIK project is an important step towards a more user friendly information retrieval environment but is by no means considered to be a final solution. While the interface has certain advantages, there are also a number of drawbacks. One drawback is that the handling of windows, which looks rather nifty in the X-Windows environment, can be a real nuisance in the MS-Windows environment. This is partly due to the fact that a 14 inch PC monitor is small compared to the usual 17 inch or larger monitors for workstations, but also due to the poor way windows can be handled in the MS-Windows environment. A second drawback is the lack of facilities for what we have called knowledge navigation above. It is not clear yet how intra database aids, like thesauri, can be supported, and the same goes for tools in helping to select (a set of) databases for searching (inter database aids). Finally, we think that ultimately the interface should allow for creating a personal information environment which enables virtual personal libraries as transparent views on a host of local and remote sources with possibilities for annotating (text, voice, still and moving images) and generating bibliographic lists and also supporting cooperative work. These demands should keep us going for some years to come. On the other hand, if Z39.50 makes a real breakthrough, developers kits for designing interfaces can be expected to come to the market, or at least there should grow a choice in, possibly better, `install and run' interfaces available as either free-, share- or payware. Another way of achieving these goals could be by enhancing existing World Wide Web browsers.

IV.C. ELISE: coloured image databases

ELISE, Electronic Library Image Services for Europe is a cooperation between Tilburg University, the De Montfort University in Leicester, England, the London Victoria and Albert Museum and IBM Scientific Centre UK. This project is sponsored by the European Commission under its Library Program. All three partners in the project are developing networked full colour image databases coupled with Z39.50 reference database servers. The standard chosen for the coloured images is JPEG.

Tilburg University Library holds a special collection for the province of North-Brabant. An important part of this collection is the topographic historical atlas, containing about 13,000 items like maps and old prints for which a reference database already exists. All these 13,000 items were photographed and stored on Kodak photo CD. From these CD's, two types of images are derived. One thumbnail for quick browsing purposes and one for viewing on the integrated desktop. At the same time the reference database is enhanced with pointers to the images. The project started in 1993 and will end early 1995 when the database will be accessible via the KWIK graphical user interface. From this user interface coloured images can be displayed by a simple click of a button, in the same manner as Elsevier articles can be browsed.

The images stored on photo CD will be used for reproductive purposes. The main advantage of this project is that very valuable (and very vulnerable) old maps and prints, will become available to users without the need for guidance of experienced library staff. If public libraries in the province of North-Brabant hook up to the Internet, this collection will also become easily available to the public at large.

IV.D. Grey Files: research papers online

A special part of the economics collection consists of research papers, so-called grey literature, from over 200 institutions around the globe. This collection is maintained and expanded by exchange of research papers from 5 series published at Tilburg University. A special reference database, Attent, containing over 10,000 references at the moment, indexes this important material. Research papers are becoming more and more so-called preprints. They are published under the responsibility of a department in order to quickly disseminate the latest results of research and to solicit peer feedback. In a sense these publications are the frontier of scientific progress, pushing articles published in scholarly journals more and more to a mere archival status. The phenomenon has grown very fast in many, but certainly not all, areas of the scholarly community. A university library which seeks to support the research process needs to collect this information. An important consideration is that authors of this literature are not so much interested in copyright but rather in being published, which make experiments with this material less legally intricate.

The Grey Files project started from the observation that most of these preprints are produced by means of word processors and/or desktop publishing software, and should therefore exist in some kind of machine readable format. Would it be possible to capture this material and make them available as files ? This idea was discussed with another university library in Maastricht, The Netherlands, and PICA. This led to an assignment to two researchers from Tilburg University and the British Library to investigate the technical and organisational issues and to give recommendations for a pilot project for the electronic dissemination of grey literature. An important aspect in this study was the relation of such a project with developments in shared cataloguing and the document delivery server technology described above.

The study recommends a pilot project based on the observation that much of what is currently being published on the Internet is of poor quality. Therefore quality assurance is an important factor. This quality depends on the consistent availability of documents and a responsibility of libraries to guarantee adequate cataloguing. Quality of the material itself is the responsibility of editorial boards of series involved.

An important issue is the format in which to make this material available. The report reviews several formats but concludes that Postscript seems to be the only viable solution at the moment. It is a widely available format for both text and graphics, extensively in use in the academic world and it can be generated from many popular word processors. Postscript also allows for full-text indexing and viewers and printer drivers are available in the public domain. Another important advantage is that Postscript allows for a 100 percent electronic duplication of printed copies of research papers. In a later stage SGML could also be considered because of its superior possibilities, such as hyperlinking with other (grey or white) documents in a World Wide Web environment. At the moment however there is not enough support for authors to warrant such an approach, although this could change if the Web develops further. Besides Postscript, ASCII versions derived from Postscript, can play a role for character based terminals.

The report considers much of the material available on the Internet poorly indexed and recommends that the grey literature should be catalogued analogously to their printed pendants. The catalogue record should be enhanced with a pointer to an electronic location, a so-called URL (Uniform Resource Locator). These URL's point to file servers where the full text files can be retrieved. These servers can be browsed via FTP, not a user friendly solution, but also via World Wide Web browsers, hiding the intricacies for the end user. Full text searching could be made possible with WAIS indexing. Another way of making these files available is via document delivery servers (see section IV A above) while a mailserver offers yet another way. Tilburg University will also migrate the Attent database for accessability via the KWIK interface which will be extended with a Postscript viewer in order to make the full text available on the integrated desktop.

The universities of Maastricht and Tilburg have started a pilot project to capture electronic versions of research papers produced at their institution and catalogue them in the Dutch union catalogue. Based on this experience the model could evolve towards a distributed system for the publication and cataloguing of preprints. Authors at both institutions, which have a strong responsibility in the project for delivering both printed and electronic versions, show a great interest and enthusiasm.

IV.E. ERA: Electronic Reference Aid

As mentioned several times before, knowledge navigation becomes an increasing problem in a world of rapidly growing possibilities for (networked) information retrieval. The challenge is to find ways of supporting users in the selection of relevant sources to satisfy their information needs. What the ERA project tries to achieve is in fact a way to record the knowledge of existing reference staff so that it can be tapped by users.

ERA consists of two parts. First of all it is a database which contains descriptions of sources, or, stated another way, it is a catalogue of secondary information, a meta-catalogue. Secondly it is an interface to this database allowing users to interactively determine the subject of their information needs. If the subject of a search is determined, the system should select sources from its database and suggest these to the user. In a next phase the system should be able to connect to electronic sources if these were suggested, or even better, start a search in such a source and return the results. This phase is not yet under development though.

ERA makes a distinction between generic and specific sources. Generic sources belong to broad subject areas and are classified with a crude classification. Specific sources on the other hand are classified using a more refined classification method. Both classification methods are coupled so that, once a subject is determined, both generic and specific sources can be recommended by the system.

At the moment ERA uses a decimal classification to determine the subject of a search. A user types in a word or phrase and the system presents the decimal code, along with its meaning, which matches the user's input in its environment, allowing the user to further determine the scope of his subject. After the subject is determined, ERA gives a number of suggestions:

Search the OPAC using a proposed decimal subject index. Eventually this action should be undertaken by the system itself.
A number of headings used for shelving, guiding the user in a physical browse of the collection. Here too, a link with the OPAC should be made possible to browse the shelf also electronically.
An enumeration of specific sources like specialized bibliographies whether or not in printed form.
An enumeration of broad subject areas which can be used to determine generic sources like CD-ROMs, online databases and the like.

In the past two years, subject specialists have been building the system and selected relevant reference material to be described and input. At the same time a crude user interface was developed. The system is scheduled to be operational in its first version by the end of 1994. User feedback will be solicited explicitly to obtain guidance in the further development of this reference aid. A second important goal is to gain insight into the maintenance cost of a system which duplicates in fact much data already stored in the OPAC. Other important aspects to be considered are integration of Internet sources and integration with the KWIK GUI.

V. Conclusion and prospects

V.A. Evaluation and success factors

After two and a half years of operation of the new library a first evaluation could be made:

The new library is a great success. Every day the building is overcrowded with mostly students who want to write a paper using one of the integrated workstations or who want to use the library in a traditional way.
The integrated desktop is campus-wide accepted as a "de facto standard" and as a key-element in the infrastructure of the university.
A development is under way to implement the use of the integrated desktop and the use of electronic information in various courses. On the other hand it is clear that it takes time to remodel courses and to optimize the integration of electronic information services and electronic communication facilities in the curricula. A technology push alone is clearly not enough.
There are also problems, of course. First of all there is a demand for more; more software, more computers, more printers, more support, but also a demand for more budget to buy books and journals.
The maintenance of the systems requires a lot of attention and a continuous effort on the part of both the library staff and the staff of the computer centre.
Because of the heavy use of the services, performance problems require attention. New solutions for new problems constantly have to be found.

The main success factors of the library innovation at Tilburg University can be summarized:

First of all, there was a vision, a strategic plan. This was the framework for various projects, in the past, at present and in the future. Because the goal was defined, various milestones could be set and accomplished. The new projects which are running at the moment fit completely in the global vision presented in the original program in 1989.
Close cooperation between the library and the computer centre was a decisive factor. Without such a cooperation library innovation will become extremely difficult and very expensive.
Commitment from the Board of Governors. The university library is part of an institution. The development of the library and a vision on the use of scholarly information should be a part of the strategic plan of the university. In that situation the management of the university logically will be committed to library innovation. At Tilburg University this has been very effective in order to get external funding and to allocate local resources.
The most important factor are the people. Staff members who work with enthusiasm and creativity to develop a new library and new services.

With respect to this last factor it must be stressed that this takes time. First of all because libraries will need staff with better and more advanced qualifications than before. This means that a library needs a long term personnel policy plan including continuous education, courses and on-the-spot-training. It also takes time because libraries are moving from rather traditional and conservative organizations into a completely new world with new goals and objectives, since users can access information without the need of the intermediate role of the library.

Already in 1984 the library deliberately introduced a new personnel policy. Library management decided to set higher requirements for new staff. At the same time a campaign was started to promote further education of the present library personnel. Staff were told that if they wanted to move ahead in the library of the future, they should follow courses, attend seminars, etc. It was regarded to be the best way to improve information services and to make jobs richer and more interesting. As a result of this policy, 36 staff members opted for further education in the period 1986-1992. Because of this policy the library could rely on its own staff in the development of new services. The experience gained in these years is a solid basis for continuous innovation in the years to come.

Another interesting aspect was that library innovation and new services did not cause a reduction of staff. On the contrary. In 1982 the number of staff was 46 fte, in 1992 54 fte and in 1994 62 fte. This increase in staff numbers can be explained as follows:

More user support in the field of information service is needed because the number of students almost doubled in twelve years.
More user support is required because of the introduction of new IT in the library; a special help-desk was established in the library.
The maintenance of existing library databases and the development of new innovations require more staff.
Various external commercial services and host activities could be started. A contract was closed with provincial authorities on the keeping of a special regional collection. Commercial contracts with publishers, business and industries were closed.

The last reason is probably the most important factor because the library funding by the university only increased with 1.5 fte In total 14.5 fte is now paid by external funding, special projects and commercial activities.

V.B Future prospects

Since the inception of the library innovation program in Tilburg it has become more and more evident that the information chain is in turmoil. The role of the traditional intermediaries, libraries, primary and secondary publishers and subscription agents is changing. At the same time scholars are becoming increasingly independent as publishers and gatherers of (full text) information on a.o. the Internet. Indeed, the very question is what, if any, should be the role of libraries in this future ?

Tilburg University library seeks to play an active role in supporting scholars and students through desktop integration and knowledge navigation. These aspects will remain central in scholarly work and perhaps increasingly so. At the same time we think that libraries should become more proactive towards their users and try to anticipate their information needs. By building collections libraries have always done so and in fact aided users in determining the usability of information. This skill is even more important in the chaotic circumstances on the Internet at present. This implies that librarians no longer wait until a user comes seeking for information but design procedures and systems to actively point users to information on a realtime basis. Information will seek users instead of the other way around. At the same time, through enhanced desktop integration users will be supported to prepare and publish their manuscripts, be it in printed through traditional channels or as electronic publishers on the Internet. An important principle is that the user should be supported wherever he is working, at the institution, at home and during trips.

Another important task we see for libraries in the future is in the training of novice users. Information literacy and the efficient handling of information will become ever more important skills. This implies a more active role for librarians in the educational process.

Collection building will remain important but will have to be balanced against access to remotely available information. The own collection should support basic needs in research and education, access to remote resources should be fast and transparent. This implies that library systems should be based on open standards which make transparent solutions possible. System vendors which use closed standards will probably not survive in the emerging complex information infrastructure.

All in all, libraries should be more dedicated to serve their users than ever before. A fine balance has to be struck between technology push and the pull from user needs. More information on user needs has to be generated, while at the same time users have to be actively informed of new possibilities.

Bibliography

Dijkstra, Joost and Ngu, A.H.H., Ariadne: a standards compliant and integrated electronic document delivery system. In: Proceedings of the 17th International Online Information Conference, London, pp. 63 - 76 (1993).

Dijkstra, Joost, Complex Objects and Versioning in Complex Environments. In: Lecture Notes in Computer Science 82: Entity Relationship Approach '93, Elmasri, R.A., Kouramajian, V. and Thalhiem, B. (eds.), pp. 13 - 23, Springer Verlag.

Dijkstra, Joost, Digital Libraries in the mid-Nineties: Ahead or on Schedule, Information Services & Use, forthcoming.

Geleijnse, Hans, The Electronic Library. In: European Research Libraries Cooperation, ERLC, pp. 153 - 159 (1993).

Geleijnse, Hans and Grootaers, Carrie (eds.), Developing the Library of the Future; The Tilburg Experience (Tilburg), Tilburg University Press (1994)

Geleijnse, Hans, Journal Articles on the Desktop : Elsevier and Tilburg Experiment, Management Information (London) 1(6) p.34-35 (1994)

de Kort, Jos and Vermeer, Ellie (eds.), Telephassa Seminar on Innovative Information Services and Information Handling : Proceedings (Tilburg) Tilburg University Library (1992)

de Kort, Jos and Vermeer, Ellie (eds.), Telephassa Workshop on Databases : Proceedings (Tilburg) Tilburg University Library (1992)

de Kort, Jos and Vermeer, Ellie (eds.), Telephassa Seminar on Information, Communication and Networking : How to Make Full Use of Available Information Resources : Proceedings (Tilburg), Tilburg University Library (1993)

de Kort, Jos and Vermeer, Ellie (eds.),Telephassa Seminar on User Support, (Inter-)National Cooperation and Strategy : Important Issues for Libraries : Proceedings (Tilburg), Tilburg University Library (1993)

The New Library and the Development of Innovative Information Services at Tilburg University (Tilburg), Tilburg University Press (1989)

Ongering, M. and Wesseling, M., Producing a Bibliographic Database Through Scanning and OCR : The Online Contents Project in the Royal Library of the Netherlands, Program (London), 26(4), p.393-399 (1992)

Roes, Hans, Current Awareness Services at Tilburg University, Electronic Library (Oxford), 11(2), p.99-103 (1993)

Roes, Hans and Dijkstra, Joost, Ariadne : The Next Generation of Electronic Document Delivery Systems, Electronic Library (Oxford), 12(1), p.13-20 (1994)

Wieërs, J.L., Planning for the Future [Part I], Excerpta Informatica (Tilburg) 3(4) p. 1-2 (1987)

Wieërs, J.L., Planning for the Future [Part II], Excerpta Informatica (Tilburg) 3(5), p. 1-2 (1988)

Wieërs, J.L., Documentation, Information and Communication at Tilburg University : Plan of Action - Research - Services (Tilburg), Tilburg University Library (1990)

Author Information

Hans Geleijnse is librarian of Tilburg University and managed the HT DIC program. He is a member of the PICA board. Hans Roes is deputy librarian at Tilburg University, managed the Online Contents project and initiated projects in electronic document delivery.

Acknowledgements

The authors wish to thank their colleagues Jola van Luyt, Thomas Place, Jos Kuijlen and Cees Zoontjes for their support and comments on earlier drafts of this paper. We remain responsible for any errors and omissions.