The Digital Library As A Personal
Knowledge Platform: LIBERATION

Von Robert Stubenrauch, Barbara Vickery und Hans-Adolf Ruppert

Within the European project LIBERATION (1,2), a group of science publishers, university libraries, and technology suppliers set up and evaluated an advanced full-text and multimedia digital library. The system is Web based and aims at adding significant value to electronic information pools. The chosen approach is not limited to enriching material with multimedia components and providing enhanced access. Instead, LIBERATION can be considered a working platform supporting knowledge management for individuals and groups by employing a range of innovative techniques such as information modularity, automated structured navigation, access control, personal views and annotations, and server clustering. The paper describes these features, some pilot applications, and outlines potential future developments.

Die digitale Bibliothek als persönliche Wissensplattform: LIBERATION

Im Rahmen des europäischen Projektes LIBERATION (1,2) erstellte eine Gruppe von Wissenschaftsverlagen, Universitätsbibliotheken und Technologie-Integratoren eine digitale Bibliothek auf Volltext- und Multimedia-Basis. Aufbauend auf Web-Technologie zielt das System darauf ab, signifikanten Mehrwert für elektronische Informations-Pools bereitzustellen. Dabei wird das Material nicht nur aufgewertet und der Zugang dazu erleichtert, sondern LIBERATION kann als eine Arbeitsplattform für Einzelpersonen und Teams angesehen werden, die Wissensmanagement im weitesten Sinn durch eine Reihe von innovativen Techniken unterstützt. Dazu gehören Informations-Modularität, automatische strukturierte Navigation, Zugangs-Kontrolle, "persönliche Sichten" und Notizen und Server-Clustering. Diese Arbeit beschreibt das System, einige Pilot-Anwendungen und mögliche zukünftige Entwicklungen.

La bibliothèque électronique comme plate-forme de connaissances personnelles: LIBERATION

Dans la cadre du projet européen LIBERATION (1,2), un groupe composé d'éditeurs scientifiques, de bibliothèques universitaires et de fournisseurs de technologie a mis au point et évalué une bibliothèque multimédia numérique à textes complets. Le système se base sur le web et a pour objectif d'ajouter une valeur significative aux centres d'information électronique. L'approche ne se limite pas à enrichir le matériel avec les composantes multimédia et à fournir un accès plus attrayant. Bien au contraire, LIBERATION peut être considéré comme une plate-forme de travail s'appuyant sur les connaissances de management de groupes et d'individus et se servant de toute une panoplie de techniques innovatrices telles que la modularité de l'information, la navigation automatique structurée, le contrôle d'accès, les vues et annotations personnelles et le groupement de serveur. L'article décrit ces caractéristiques, quelques applications pilotes et souligne les développements potentiels à l'avenir.

1. Introduction

Global unlimited access to the full body of information available to mankind has been a wish and a vision for ages. The rudimentary realisation of this dream has until recently been a privilege of a growing, however still small number of scholars. With the advent of information systems based on global computer networks such as the Internet which is considered the largest information repository mankind has ever had access to, in the last few years this dream has moved into the realm of possibility. Especially in the academic world, with the help of computer technology it no longer seems impossible to connect existing information bases and build up a global network of scientific and educational information available to the general public. Digital libraries is one term that is used to talk about such systems, but it is quite a vague term which can mean at least three different things:

• Online library catalog: Remote access to an electronic library catalog, providing information which paper based publications are physically available in a traditional library and hints where to find them; such systems are called OPACs (Online Public Access Catalog).
• Digitised archives: Access to digitised collections of real-world objects, such as manuscripts. In that way access to materials can be offered that would otherwise be prohibited for conservatory or security reasons.
• Full-text and multimedia material: Material originally prepared in digital format which allows full exploitation of the possibilities offered by such types of services. Such systems are sometimes referred to as "virtual libraries".

LIBERATION focused on the last aspect. Drawing on extensive experiences with videotext systems, computer based training, Intranet and Web applications, and electronic publishing a consortium of publishers, users and technology suppliers has undertaken to specify and implement the prototype of a comprehensive working platform based on a repository of digital full-text and multimedia material. The basic approach of the system is to add as much value to electronic publications as possible by fully exploiting the sophisticated structuring and knowledge management features provided by the underlying information management system. In that way all affected groups of users, from the production end through distributors to the consumers’ end, would benefit as much as possible from using such a system.

2. The vision

LIBERATION is specifically aiming at university settings which means that its content mainly comprises scientific papers, proceedings and textbooks as well as other educational material at academic level. While this environment does come with some particular requirements the scenario of usage may easily be extended to general Intranets of large research institutions or other organisations that rely on flexible, structured management of large volumes of interlinked, but unstructured information. In fact, the number of potential users of such a system is rapidly growing these days as global enterprises are having to cope with a dramatically decreasing half-life of knowledge.

With this challenging background, in the design phase of the project, a number of basic criteria were identified as pre-requisites for a LIBERATION type digital library:

1. Compatibility:
   Full compatibility with current WWW technologies and preparedness for future developments;
2. Data formats:
   Support of a wide range of text and multimedia formats;
3. Search:
   Sophisticated search options including search for title, full-text and meta-data;
4. User interface:
   A consistent graphical user interface across publications which would result in an increased general usability and user acceptance;
5. Content individualisation:
   Possibility to let the user put material into different contexts which requires a high degree of modularity of the material;
6. Personal work place:
   Users should be able to maintain a private "area" in the system which they can partly open to (groups of) identified users;
7. Annotation:
   Possibility to the let users annotate any visible item of information, regardless whether the user has write access to the object itself or not. Visibility/accessibility of annotations should be controlled by hierarchic user rights;
8. Separate content and graphical user interface (GUI):
   High degree of separation between the user interface and navigation features on the one hand and the content on the other hand;
9. Access management:
   Possibility to present material and links (!) depending on hierarchic access rights;
10. Server clusters:
    Possibility to cluster servers into groups which share the same user accounts and groups and allow to search across server boundaries;
11. Meta-data:
    Flexible handling of meta-data in conformity to current standards;
12. Mass data maintenance:
    Efficient tools for mass data maintenance must be available;
13. Billing:
    Support of various billing and charging models should be provided

In general, the project consortium envisaged a digital library that should be more than just an information repository offering access to information. Rather, it should provide a working platform for knowledge workers which would allow groups or individuals to explore and manipulate large bodies of interlinked information. A main part of such work naturally consists of creating new information materials or compiling existing ones into new views. Thus the roles of information producers and consumers would blend in such a system.

3. The Underlying Concept

It would have taken much too much effort and time (considering the extreme short development cycle in that area) to build a system from scratch meeting all these requirements. Therefore the decision was to base LIBERATION on an existing information management system that would offer enough flexibility to be customised into the desired direction. The system of choice was Hyperwave (3) (then being a non-commercial prototype code-named Hyper-G), a Web based knowledge management system with a range of uncommon, very useful features that makes it an excellent basis for very large, highly customised Intranet solutions. That this choice was right is indicated by the fact that Hyperwave won the Byte Magazine' "Best of the Show" award at CeBIT 97 as well as the European IT Prize in November 1997. Another important factor speaking for Hyperwave was that it is free of charge for educational institutions, an essential fact considering that the users within LIBERATION are university libraries and institutes.

3.1 Hierarchical Structure of Content

Based on Hyperwave, LIBERATION comes with the concept of content, structured hierarchically in a specific type of folders. In contrast to file systems, in LIBERATION objects can be available logically in multiple folders and the folders that contain an object (the "parents") can be easily addressed from the object itself. This is possible because of the powerful link handling possibilities of LIBERATION: all links are bi-directional and kept separately from the content. This is in strong contrast to regular Web servers where links are embedded in the documents themselves, for instance via the anchor tag in HTML. The frustrating broken link experience of plain HTML Webs is avoided in LIBERATION because link consistency is maintained automatically.

3.2 Document Attributes

Just as links are kept separately from documents, so are a range of other attributes. Meta-data of any type can be attached to data objects in LIBERATION. Some of them are defined by the underlying system. For instance, when the user opens a folder with the attribute "sequence" set, the system dynamically links together all objects within that folder (the sort order is another attribute), and depicts them appropriately. This is an extremely useful feature that drastically reduces maintenance effort whenever documents are deleted from or inserted into sorted collections of documents.

Another attribute enables the identification of one specific document out of a folder as its "head" so that when the user opens the folder, actually its head is opened, optionally followed by a list of the remaining content of the folder. An attribute called "multicluster" allows all documents in a folder to be "glued together" so that they appear as a single document. Yet another attribute is called "name" and represents a character string unique on the server which is used to access documents through the WWW in the same way URLs are. However, in contrast to plain Web servers, LIBERATION allows multiple names or URLs to be assigned to objects.

As names in LIBERATION have no relation to the physical storage of documents, moving objects between folders or logically "copying" them to further folders (which is in fact just linking), does not affect their accessibility through their URLs. This again is in contrast to traditional Web servers where a URL corresponds to the actual path where the document is physically stored. Of course another attribute allows to assign meaningful titles to objects, even in multiple languages.

3.3 Meta-Data

In addition to these basic attributes, further ones can be freely defined and used by means of customisation. This feature is used to provide publication related meta-data conforming the Dublin Core schema (4,5,6). In the text below a sample listing of attributes is given as it is embedded in the HTML served by the system. It shows LIBERATION internal attributes as well as Dublin Core meta-data, recognizable by the prefix DC. Note that the WWW end of the system could be configured to produce any other standard of meta-data as well; Dublin Core is used in the pilot phase to demonstrate this possibility.

In a typical LIBERATION document Web (Fig. 1), documents of various formats are structured into folders. The thin solid arrows represent structural relations between objects ("is contained in", "is followed by" etc.) that are dynamically transformed into corresponding links when the documents are retrieved. In most cases it is actually possible - and desirable - to build large Webs without any "hard-coded" referential links but just on the basis of such structure elements. However, in our example there are two traditional cross reference links (hyperlinks) represented by the thick dotted lines.

When a plain unstructured Web of documents is imported to LIBERATION, which is the common way to add material, structure elements are automatically created following the folder and name structure of the original material on file system. Although the material can be used without any modifications in LIBERATION as it was on the original plain Web, by adding proper LIBERATION attributes and restructuring the Web a bit into folders, significant value can often be added to existing material in an elegant and efficient way.

Because objects can have multiple "parents" it is possible to create different "views" of the same material by putting it into different folders. This is depicted (Fig. 2), where, depending on where the user "enters" the Web, the same documents are accessed through different paths. Additionally to this structuring option, true "personalisation" of material can be achieved by applying appropriate access rights. In LIBERATION, various kinds of access rights (read/write/unlink rights) can be assigned to any object and user groups can be structured hierarchically similar to how user accounts in UNIX are handled. In that way access rights can be assigned group-wise or user specific. This is an important feature considering applications such as online lectures in which access should be provided only for class members. Only the lecturer should have access to the students’ homework but not their fellow students.

So, to sum up what has been said so far, LIBERATION is based on a hierarchically structured information management system. The hierarchic structure of content allows provision of a range of useful navigation facilities and results in a reduction of maintenance effort and in an improved consistency. The structure also permits a modularisation of the material which is a precondition of increased reusability, putting material into various contexts. Attaching arbitrary attributes to any type of object allows current requirements regarding meta-data to be met. LIBERATION provides sophisticated search options including search for meta-data. Because LIBERATION is based on a general and open knowledge management system, it can easily be integrated with modules for Web based training, computer supported collaboration (CSCW) or other components further increasing its knowledge management power. For instance, a feature that is currently not used in LIBERATION is versioning which might not be necessary in the context of university libraries but can be essential for rapidly evolving libraries of large corporations.

4. The system

LIBERATION is an open distributed digital library system, based on Internet technology and WWW in particular. The underlying information management system Hyperwave, and thus LIBERATION itself is available on all common UNIX platforms as well as for Windows NT. The features of the system can be fully exploited with any 4^th generation Web browser; no further specific hardware or software is required on client side. LIBERATION servers can be clustered to build a consistent group of servers that allows shared access to parts of the content and a central administration of user accounts.

4.1 Physical Structure

The physical structure of a LIBERATION cluster is shown in (Fig. 3): Embedded in the global Internet and accessed through its most popular user interface, the WWW, there are a number of servers running within LANs, for instance maintained by libraries. The clustered servers are communicating directly through a dedicated server-to-server protocol in order to provide consistent functionality across server boundaries. They are accessed locally from within the respective LANs or from anywhere on the Internet. Basically there are three modes of access, depending on the location (the Internet domain) from which a user has registered.

Accessibility of Features and Content

Because some features do not make sense for unidentified users (home area, annotations, edit etc.) and access to some material might be restricted for commercial or legal reasons, features and access are available in three modes according to Table 1. Often, unidentified users may access title pages and tables of content of restricted publications so that they at least "know what they are missing".

Table 1: Feature and access availability

4.2 Logical Structure

The logical structure of a LIBERATION cluster, i.e. the content as it is perceived by users, looks quite different, as depicted in (Fig. 4). In this example there are three Hyperwave servers on top that may contain material beyond LIBERATION. At some level in the folder hierarchy there are local entry points to the LIBERATION library. Where the content, represented by the triangles, intersects, the documents are available on all the intersecting servers. In particular this means that searches can be performed across server boundaries. Still, local administrators of the individual servers can structure the material according to their specific needs, for instance along certain classification schemes or just by author name. Because documents can be contained in an arbitrary number of folders, structuring in more than one way is possible and often useful.

As a further benefit of server clusters, user account management may be centralised in a shared user pool so that user accounts on one server are valid under the same conditions (such as group assignment) on the other servers of the cluster. This reduces maintenance effort significantly, presuming that within a digital library cluster, access conditions are homogenous due to overall licence agreements.

For the users the fact that more than one server is involved is irrelevant and may even not be noticeable. Because the servers communicate with each other directly, the URL that the browser shows in its status line is always the URL of the server to which the user has connected first in the current session.

4.3 Functions and User Interface

The user interface of LIBERATION is highly configurable through a script language that features system specific modules embedded in JavaScript code. These scripts allow access to server status parameters as well as content specific meta-data and present them in an intelligent and attractively laid out way.

Screen Design

A screen shot of a typical title page of an electronic text book in LIBERATION is shown in (Fig. 5). The screen is divided into three basic areas which are generated by use of HTML frame technique and are consistently presented throughout an entire session. Area A holds the actual content of the current object. In our case it is the title page of a book entitled "Hyperwave - The Next Generation Web Solution" by Hermann Maurer. While in our example the entire publication is available in HTML format, please keep in mind that arbitrary document formats can be stored in LIBERATION. Of course each format may require specific software components to be displayed appropriately. However, "plug-ins" for popular Web browsers that display common formats such as PDF or multimedia are usually available for free download on the WWW.

While the content in area A is a "hard-coded" part of the actual publication and its appearance in LIBERATION can therefore not be determined by the system, areas B and C are dynamically generated by LIBERATION and feature general system functions and content related navigation options.

Let us have a closer look at C first. In this area the dynamically generated navigation features are presented that implicitly result from the structure of the documents and their attributes. The main part is a listing of the content of the current folder; in the case of a text book this corresponds to the first level of the table of content (display as tree structure would also be possible by customisation). The height of area C is automatically adapted according to the number of elements the current folder contains.

Parent Folders

On the top of area C there is a drop-down menu that allows users to select the required parent folder. In the example shown in(Fig. 6) the current document is contained in two folders, one of which is the "home collection" of a user named Miller. This might be a folder that is only accessible for this single user. The system name of the user currently logged in is shown in the left lower corner of the screen.

Sequences

In (Fig. 7) another helpful component of area C is shown: sequence buttons that allow access to all documents of a folder sequentially (remember the "sequence" attribute described above). When the user points the mouse on a button, the title of the document is momentarily displayed allowing a simple "preview" of what comes next. Also, the position of the current document within the sequence is shown (6 of 7).

Let us now turn to area B (Fig. 8). Here, important system features are available, including a jump to the libraries entry page, search and info (help and news on the system and newly arrived publications, contact information etc.). "Preferences" allows definition of the quality of the documents to be shown and the level of detail of the meta-data to be listed. "Attributes" list the meta-data of the current object and "Title page" is a short-cut to the title page of the current publication. This is particularly useful if the user does not access a document by going down the hierarchy from the title page, but accesses it directly, for instance as result of a search query. In that situation, without the context of the document, the user might be lost otherwise.

Language Switching

Another feature available in this area is language switching. Currently the user interface is available in English and German but more languages could easily be added. Switching language means that the entire user interface is presented in the selected language and the content (if it is available in that language of course).

You may have noticed that some of the buttons are shown in dark. These are functions that are not available for unidentified users, including "Home" (jump to personal home folder), "Annotate", "Edit" and "Copy". If the user tries to select such a button she is requested to first log in.

Document Attributes (Meta-Data)

The "Attributes" button displays a selection of the attributes of the current object. (Fig. 9) shows how this feature would depict the attributes of an object similar to the one whose meta-data are listed in Table 1 above. Presence of attributes might vary from document to document as data are available.

Annotations

Another quite essential feature hides itself behind the three dots next to the "Annotate" button. If the user points the mouse over this little button the result might be as shown in (Fig. 10), displaying the number of available annotations. If the button is clicked a list of the titles of annotations attached to the current object is shown in the same style as folder content is listed.

Identified users can annotate any object they can read, even if they do not have write access to the document itself. This again is possible because links are kept separate from the content. The texts to which the annotations actually link can be put anywhere on the server where the user has write access to. By default this is a specific subfolder of the user’s home folder that is automatically generated upon registration.

Note that annotations can in turn be annotated themselves; in that way simple discussions can be built up. However, we are aware that the annotation feature of LIBERATION should be carefully redesigned to introduce even more powerful features such as

• Annotating parts of documents;
• Annotations as links to other documents;
• Annotating groups of documents;
• Ability to annotate in form of up-loaded documents of any format;
• Annotation icons directly embedded in the annotated document;
• Typed annotations; types could be "question", "pro", "con" etc;
• Intelligent handling of modifications of annotated documents.

All these features are well feasible within the framework of LIBERATION because of its flexible basis.

Edit Mode

Switching to the "Edit" mode takes the user to a completely different user interface, the "power user mode". This allows exploitation of all the features of the underlying Hyperwave system without restrictions. This mode is considered useful only for users who want to add or maintain material on the server. All maintenance work can be done via standard Web browsers, including user account management, which, however, is restricted to system administrators. While some of the functions are naturally a bit complex, a range of manuals and online help are available and provide professional support. For uploading and maintenance of mass data (some of the publications in LIBERATION such as dictionaries consist of many thousands of individual documents) there are command line tools available for the experienced professional user.

Copy

A simple but effective and often sufficient alternative to the powerful "Edit" mode is the "Copy" feature. Clicking on it allows the user to add the current object - regardless whether it is a folder or a single document - to another folder. The default destination is the user’s home folder so she can easily collect material that is of personal interest to her, quite similar to bookmarks known from Web browsers. The difference and added value in LIBERATION is that - because links are bi-directional - the bookmarked object points back to the folder in which it is contained via the parent link (which, however, might only be visible for the user who set the bookmark).

Search

Of course a crucial feature of any digital library is a powerful search mechanism. In LIBERATION the following search related features are available:

• Search across server boundaries in server clusters
• Search can be restricted to specific folders
• Full-text search in nearly all known text formats
• Title and attribute search
• Iterative search
• Sort search results by various criteria
• Logical combinations of search expressions

An interesting possibility of LIBERATION is to create "query" objects. These are objects containing all parameters of a search query as they might be entered by a user. When opening a query object the search is performed and the results are listed as usual. By the use of this object, performing frequent queries with the same parameters is made very comfortable. It can even be automated by setting a time parameter (daily, weekly etc.) at which the system performs the query and sends the result via e-mail to a predefined list of recipients. This is a first step towards the direction of "personal agents" that will be one area to explore deeper in the near future.

Examples of a search dialog in LIBERATION and the resulting list of objects are shown in (Fig. 11) and (Fig. 12).

5. The Project

Having described the system in some detail, let us now have a look at the project itself. LIBERATION was a project funded by the European Commission, DG XIII (Telematics Applications, Libraries Sector) and ran from May 1996 to November 1998.

The Project Consortium

Regarding the consortium, three groups of partners can be distinguished: first there were three German publishers (Springer, Addison-Wesley, BIFAB) who contributed material and their points of view of commercially interesting issues such as billing and copyright. The publishers were the main content providers in the project although significant amounts of material were contributed from other sources.

Then there was a group of three user organisations, including Freiburg University Library (Germany), Nottingham University Library (UK) and the IICM, an institute of the Technical University Graz (Austria) that plays a major role in academic education in the area of computer science and applications in Graz. These partners specified the requirements on the system in an early stage of the project and were later using it in local pilot applications. They also carried out thorough evaluations of the system and its usage.

The IICM as the co-ordinator of the project was also a member of the third group of partners, which is the group of system integrators and technology suppliers, where it was joined by the IfI, an informatics institute at the University of Freiburg (Germany). The task of these partners was to customise the system in order to meet the specified requirements, to provide the tools for evaluations, to assist the user organisations in setting up and maintaining the systems and the content providers in preparing materials.

The overall aim of the project was to provide a prototype of an innovative type of digital library that would meet specific requirements of all involved user groups, to use it in pilot projects, and to evaluate it in order to draw conclusions for future work.

6. The Content

Content-wise, the main focus in LIBERATION was on academic literature in the area of computer science and applications to which publishers contributed from within the consortium and from outside. Besides a standard computer science dictionary (DUDEN Informatik), digital versions of text books on algorithm design, computer supported co-operation, Internet and the WWW, and Web access and programming (Java, Web browsers, Intranet building etc) were available. Additionally, a number of conference proceedings on Web issues and computer-based training ware provided (ED-MEDIA and WebNet conference series) as well as local collections of diploma and doctoral theses. Most of these materials are available in HTML format, some in PDF. A set of courses on various aspects of data-bases was made available based on HTML with integrated Java modules supporting interactivity.

In addition to these thematically focused materials, further valuable publications of general interest were made available: a fully structured and cross-referenced bible (Elberfelder Bibel) and a general German/English pictorial dictionary (DUDEN Bildwörterbuch, see Fig. 13). Locally, at the three pilot test sites, further substantial material was created and used in the framework of pilot applications (see below). The overall offer was rounded up by links to other full-text and multimedia material available on Hyperwave servers such as Web based computer journals (J.UCS, JNCA) and an English/German dictionary (Langenscheidt). The material is currently structured hierarchically by the type. The entry page of a typical LIBERATION system is shown in (Fig. 14); it lists the categories of publications that are available.

Although three large publishers were members of the project consortium, and even though the focus was on a small sector regarding the subject area, namely computer science and applications, the problem of reaching a critical mass of high-quality content was still evident in LIBERATION, as it is in most similar projects. In LIBERATION, because it is a European project, there was the further issue of language. While it is acceptable for most German speaking students to read English literature (particularly if it is computer literature) this is not the case vice versa.

Volume-wise the shared digital library of the three sites at Nottingham, Freiburg and Graz contains about 60 volumes of publications, where the exact number depends on how individual documents are counted. The number of individual "atomic" documents (such as individually accessible HTML pages) is about 200.000.

Business Issues

Within the pilot applications of LIBERATION there is no charging of the users of the system. Some publishers have provided their material without any restrictions, others have formally restricted the usage for the time the project is running and for identified access. However, the system is prepared to provide a range of various charging and billing options from which publishers may choose. If the system is turned into a commercial service this will certainly become a crucial issue. Publishers will have to develop innovative models for producing and distributing electronic material just as system integrators will have to turn these concepts into reality. All in all publishers will have to radically rethink business with new media. They will start to consider value that is added by the underlying systems an integral part of their products. The content itself is only part of the value; the way content is connected and can be used individually may be the value that is even more appreciated; the relations between the documents and between parts of the documents provide an exploitable multiplier effect. The manifold features provided by LIBERATION are a good example for this concept.

7. Pilot Applications

In what follows we give a brief overview of the pilot applications that were carried out at the test sites in the course of the project.

7.1 University of Freiburg Library, Germany

In Freiburg, the development of project LIBERATION was marked by the technical organisation, contextual enhancements and the integration of the system into the existing electronic library services. The focus of the subjects to be covered was basically on computer sciences, physics, as well as theology and philosophy. Additionally, an important and comprehensive collection of multimedia material was compiled for the studies in German philology. The advantages of the Hyperwave system in contrast to conventional WWW server systems have quickly been realised and used by the contributing authors. Thus the modern system resources provided were used wherever possible and useful and a "digital library" was created in a relatively short period of time, serving the specific needs of the University of Freiburg which is rather Arts and Humanities oriented.

Another main working area was centered around the task of integrating the LIBERATION system into the general electronic services provided by the library. Most of the electronic documents supplied through LIBERATION cannot be kept in the online public access catalogs, mostly due to official German cataloging rules. The necessary information is therefore supplied through an extra online catalog dubbed "EP-Navigator". The EP-Navigator can easily be accessed via the World-Wide-Web. User identification is a management problem still not solved sufficiently: On the one hand, the system itself supplies authentication procedures but on the other hand the authentication system of the university - where practically everyone eligible has already been registered - also offers the possibility of showing to what faculty the user belongs. It is therefore a future task to link external authentication servers, without creating extra administrative expenses. At the moment this situation does not present a problem, for each user must supply a separately kept password with his first login.

The LIBERATION system is offered for use in the University Library's Information Center at twelve PCs. The Information Center offers access mainly to online information from bibliographic and other scientific databases, but also gives access to some full-text databases. LIBERATION thus substantially adds to what can be offered to the user. In the very near future, additional PCs will be installed in the reading rooms of the library, also offering access to LIBERATION. Presently, a vast amount of these terminals already exist in the libraries associated with the participating faculties.

Selected university teachers offer courses in the use of the system in their lectures, started in summer semester of 1998. Unfortunately, it is hard to give well-substantiated statements about user behaviour because there are simply not enough users actually working with the system at the time. Nevertheless two trends can be made out:

• The well-organized data structure in conjunction with the excellent communication characteristics of LIBERATION makes it an ideal tool for research teams working together at different locations. Consistent data management achieved by constant online updating and the range of annotation options are much credited features beyond the possibilities otherwise offered by the WWW.
• It is already evident, that the technical infrastructure the beginner has access to - even in the more science oriented fields - is not yet as commonplace as it should be. Not everyone - in a group or alone, or even from the dormitory - can have access to the system at any time needed. Of course, the question to what extent this is the result of an unsatisfactory infrastructure, and to what extent other elements, such as a lack of general "computer literacy", low motivation to acquire new learning techniques etc. play a role in this issue, cannot be answered unequivocally at the moment.

7.2 University of Nottingham Library, UK

At Nottingham the developments within LIBERATION centered around user needs, design of statistical parameters and evaluation of system use. Apart from the materials provided by the publishers (mainly computer sciences and physics), considerable local materials were provided in other areas, in particular education and engineering which have served to enhance the relevance of the LIBERATION database for local students.

However it was recognised at an early stage that , if it was desired to fully exploit Hyperwave teaching and learning support facilities, more was needed to do than simply increase the volume of material in the LIBERATION repository. Consequently the Nottingham team sought to interest academic staff in using LIBERATION as an integral part of their teaching and results have been encouraging.

Current teaching uses of the system therefore includes:

• Ergonomics: A module has been designed as a self-paced learning package to assist in the design and assessment of work environments. Also included is an Authorware diagnostic tool for workstation design called RULA (rapid upper limb assessment).
• University Learning Support Material: Currently still under development this will incorporate a selection of study skills material for use by students who have impaired learning abilities.
• Schools Online: Similarly still under development, this project is managed on behalf of the Teaching Enhancement Office. It exploits the use of personal and shared folders available in LIBERATION to enable postgraduate students collectively to develop an evaluation tool for school use of the Internet by analysing schools' web pages.

LIBERATION is offered for use in both library and general settings. Dedicated machines are available in two major site libraries where students can gain access to expert assistance. Users may also choose to access from any other campus machine and simply make use of the support documentation and on screen assistance.

With a view to supporting the process of evaluation a registration system was designed to include a mandatory questionnaire which provides specific details about status, department etc. On completion of the questionnaire the new user is automatically assigned a personal working area which only he has rights to. He is also assigned to a user group containing all local students. Once a user has accessed the materials in the system he may then fill out the feedback evaluation questionnaire.

Unfortunately usage data has been small due to the lack of a critical mass of widely relevant materials. However some conclusions may still be drawn even from a relatively small field. In summer 1998 there were 120 registered users of LIBERATION (28 staff, 50 postgraduates and 42 undergraduates drawn predominantly from engineering and education faculties). Questionnaire responses from early trials have included:

• easy to use
• response time adequate
• good accessibility
• minimal help required
• most users printed material because they prefer to read hard copy
• none made use of their private working area

It was clear from this early feedback that minimal use was made of the full functionality of LIBERATION. Consequently further specific groups were targeted and the level of instruction they received on using the system was increased - in particular its interactive features.

Although user groups are small in Nottingham there are academics who use the system to support their teaching. It is intended to generate sufficient interest in the local university community to ensure the widespread integration of LIBERATION as a teaching and learning support tool.

7.3 IICM, Austria

At the IICM (Institute for Information Processing and Computer Supported New Media, TU Graz), LIBERATION was used in the summer semester 1998 in the framework of a seminar entitled "Social Aspects of Telematics". The class of about 120 students dealt with nine topics, each presented live by a lecturer and represented by a dedicated folder in the LIBERATION system. The students were requested to discuss the topics online by using the annotation facility of LIBERATION. In that way, some 400 documents were created (the entry page to the collection can be found at http://www.iicm.edu/aspekte98). Students self-registered on the LIBERATION system through the online registration form made available. In this way registered users automatically received the following:

• an account that allowed to access the entire LIBERATION system,
• a private working area with the standard LIBERATION layout and functionality,
• a private sub-area for their annotations.

Due to the discussion oriented nature of the seminar, some of the key features of LIBERATION were not employed in this pilot application. In particular, the creation of private material and compilation of "personal views" of existing material was not actively encouraged. Therefore, is was not expected to obtain relevant feedback on such features by the users in this seminar. However, informal talks with tutors, system administrators and users resulted in valuable insights about the user behaviour of various user groups, corresponding requirements and desirable adaptations of LIBERATION, particularly regarding the annotation feature.

Because LIBERATION has not been designed as a full interactive discussion forum it would be hard to compete with the well-known news services in this area. So let us just pick out two concrete requests for improvements: The tutor complained about the impossibility to easily annotate with the same text more than one documents, i.e. to simultaneously give a group of students the same reply instead of replying to all of them individually. On the other hand, student experienced problems when documents that they had annotated were modified by their authors after the annotations had been attached. In some cases, documents were even deleted after they had been annotated. To overcome this problem, there should be a mechanism to "freeze" annotated documents, similar to the way common "versioning" works.

8. Outlook and Conclusions

LIBERATION as a full-text multimedia digital library has gone a long way towards the vision outlined in the beginning of this paper but there is still very much work left. In general, the direction of future efforts will be to build a digital library that is extended to a true knowledge management system. After all, the purpose of libraries has always been to be a repository of knowledge. With current technology this basic quality to allow access to information can be extended to systems that really support the process of gaining knowledge - which, after all, only happens in human brains. Some future technical work that derives from these considerations might be focused on the following keywords:

• Integration with OPACs (Online Public Access Catalog);
• Integration with digitised archives;
• Incorporation of dynamic background libraries and "personal agents";
• Incorporation of content rating and filtering systems (such as he PICS standard) in order to face the challenge of the growing amount of irrelevant material on the public Internet;
• Full integration of computer supported synchronous and asynchronous communication;
• Incorporation of didactics modules and so-called Web based training systems (WBT);
• Enhanced models to dynamically add value to electronic publications while reducing preparation/conversion efforts;
• Developing new business models for new media products and implementing them;
• Developing models for self-publishing and sharing content, including the review process (especially in the academic world);
• Exploiting remote information resources, i.e. information that is not available on the local server.

Based on the experiences gained in the course of LIBERATION we can make the following observations and overall conclusions for future efforts in this field:

1. A digital library of the LIBERATION type can add enormous value to any publication that is available in the system. This however requires that the material conforms to some structural guidelines.
2. In order to ensure this structure conformity, technical tools must be ready to allow (semi)automatic end preparation of the material. It is also necessary to support proper infrastructures for human editing processes.
3. Publishers and libraries must work out business models based on the idea of exploiting the overall added value of such a system. That for instance means they must appreciate the fact that their material is (re)used in arbitrary contexts and that they do not loose profit if the user turns to a publication provided by another publisher. This requires radical rethinking and is considered a key issue. Of course this also requires appropriate technical support, in particular concerning innovative charging mechanisms. And it may require some innovative approach to the legal issues of licensing and copyright.
4. With all this measures taken it still requires very high efforts to establish and maintain a critical mass of high-quality up-to-date content that would be necessary to attract sufficient users. This refers to "original material", locally available on the digital library server.
5. Therefore, in addition to preparing and giving access to local material the strategy must be to extend the added value approach as far as possible to remote material, i.e. material that is not "owned" by the library but which can nevertheless be exploited to some degree. Depending on how much the "owners" of the remote material co-operate (for instance by applying meta-data standards) a Hyperwave based system could locally store pointers to such content, enriched with various information and corresponding functions. It could even be possible to "copy" parts of remote material into private areas and restructured it to some extent.

Finally, it must be an overall principle to listen as close as possible to the requirements of all involved groups of people and to take care that the benefits of the new digital technologies are made available as widely as possible, in particular to underprivileged segments of the population.

Above issues are addressed by the concept of a European follow-up project to LIBERATION called OiSEE (read "Oh, I see!", for Open Integrated System for Emerging Expertise) (7) that will most probably be co-ordinated by a university library. It is intended to submit a project proposal with the first wave of calls by the European Commission within the scope of the 5^th framework program in the first half of 1999. OiSEE is in an early stage of conception and potential partners are encouraged to join in during the formative phase.

References

(8) GLOBAL INFO: Germany’s Digital Library Programme: http://www.global-info.org/

Angaben zu den Autoren

	Dipl.-Ing. Dr. Robert Stubenrauch JOANNEUM RESEARCH, Institute for HyperMedia Systems Schiesstattgasse 4A A-8010 Graz Austria E-Mail: Robert.Stubenrauch@joanneum.ac.at Homepage: http://www.iicm.edu/rstubenr
	Barbara Vickery Nottingham University Library University Park Nottingham NG7 2RD UK Email: Barbara.Vickery@nottingham.ac.uk
	Dipl. Phys. Hans-Adolf Ruppert Freiburg University Library Werthmannplatz 2 D-79098 Freiburg i. B. Germany Email: ruppert@ub.uni-freiburg.de