Guido Goedemé
Royal Library Albert I
Brussels, Belgium
guido.goedeme@kbr.be
guido@lib.kbr.be
On the national level, projects include the Union Catalogue of Books in scientific libraries managed by the head librarians of the collaborating libraries, the Union Catalogue of Periodicals managed by the University of Antwerp, Belnet (Belgian Network), the network for the academic and scientific institutes with connections to the Internet and managed by the Ministry of Scientific Policy, as well as Virlib (Virtual Library), a collaborative electronic document delivery system of the Antwerp and Brussels universities and financed by the Ministry of Scientific Policy.
On the European level, the collaborative projects include the MORE (Marc Optical Recognition) project for the Intelligent Character Recognition of library catalogues; the COBRA-Metric project for the development of a bibliometric database of bibliometric data; the AUTHOR-project for the development of a common authority file for author names; the EROMM project for the development of a database of records, relating to microform masters; and the ISDS-database in Paris for periodicals
The paper also discusses how the Royal Library Albert I has to modify its GEAC's VUBIS system constantly in order to carry out the collaborative projects.
The Royal Library Albert I uses the VUBIS system for its automation projects. VUBIS is the property of the Vrije Universiteit Brussel and Technische Hogeschool Eindhoven and commercialized by GEAC. It is an integrated real time library system, based on and written in MUMPS. MUMPS, or M, as it is called now, is at the same time an operating system, a data base system and a programming language. VUBIS, although developed by academic libraries, is mostly used in public libraries in Belgium, the Netherlands and France.
In order to adapt VUBIS to the requirements of the Royal Library, a new internal format similar to UNIMARC has been defined. This internal format is based on a data dictionary, which allows for an easy definition of new tags and subfields, control of inputted tags and subfields. In the input module, frequently occurring tags are presented to the cataloger. Since different library subsections use different tags, in order to avoid the unnecessary presentation of tags, user profiles have been defined by grouping frequently used and less frequently used tags. These user profiles take into account the type of publication, the date of publication, and the subsection in the Royal Library, such as National Bibliography, Manuscripts, Rare Books, and Official Publications.
The original VUBIS-format, although not as good as the UNIMARC-format, has been kept online also, and there is a real-time conversion of UNIMARC-format to VUBIS-format. Depending on the needs, one of the formats is used for outputting data. Because M as a programming language is based on an interpreter, the Royal Library disposes on the source code of the online programs. These possibilities, as well as the fact that definition and protection of new functionalities are very well managed by the system, allow for the creation of data output files, depending on mutual agreement or user requirements.
The selection of data to be outputted is possible with the user-added programs, since VUBIS-extraction programs could not meet our requirements. The Royal Library added more extensive exemplar dependent information to the system than foreseen in the VUBIS system. This information, including book number, date of entry of the copy and a fund indicator give extra possibilities for extracting data. The result of an extraction process is kept within a so called "save- list", and this is a standard VUBIS-feature. Maintenance of save-lists, i.e. controlling, copying, fusion or deletion is performed by standard VUBIS-software. These save-lists save as an input file for the output programs.
The creation of interfaces for the off-line loading of external databases is not a problem either. Background databases, based on data of Library of Congress and Bibliothèque Nationale de France are set up for copying data into the local database as well as for searching bibliographic data or subject indexes. A bibliographic database of analytical entries of articles in periodicals, bought at Swets & Zeitlinger is set up for nearly 2500 different periodicals. Off-line loading into the local database is based on the UNIMARC-format; this facility will be used while loading data for the retroconversion project, which is a spin-off of the MORE project as we will discuss in this paper later.
Output files are or will be created for various collaborative national, European and international projects.
2. NATIONAL COLLABORATIVE PROJECTS
On the national level the Royal Library has several collaborative projects. These include:
2.1. The CCB (Centrale Catalogus van België)
Project
CCB is a joint project of academic and other scientific
and special libraries in Belgium (CCB, 1993). Both CCB I, published in
1989, and CCB II in 1993 are results of an initiative of the "Nationale
Conferentie van Universitaire Hoofdbibliothecarissen" (National Conference
of Head Librarians of University Libraries). It contains monographs in
Belgian scientific libraries. The project is managed by the universities
of Gent and Leuven. The participating institutions send their data to the
project team; contradictory to a general expectation, the exchange format
is not an ISO 2709-like MARC format, but data are sent in SGML format.
A number of reasons are cited (CCB, 1993), and the major ones can be briefly
summarized as follows:
• the greater part of the participating institutions did not handle the MARC-structure with tags, subfield codes and indicators and did not dispose of a MARC output format,
• the SGML was kept flexible and open, so that all gradations of catalography, ranging from poor to rich could be treated.
2.2. ANTILOPE (Antwerpse Index van Lopende Periodieken)
ANTILOPE is an initiative of the University of Antwerp.
It is a union catalog of periodicals and holds 150,000 titles for 164 participating
institutions. The contribution of the Royal Library is of the order of
7,000 periodicals.
ANTILOPE and CCB are the basis for IMPALA (Instant Mailing Procedure for Automated Lending Activities), a document ordering system, maintained by the University of Antwerp.
3. EUROPEAN COLLABORATIVE PROJECTS
On the European level, the Royal Library also has various collaborative projects:
3.1. The MORE Project (http://www2.echo.lu/libraries/en/projects/more.html)
(Lupovici, 1994; Vanderpijpen, 1994)
MORE has been selected through the first call of
proposals in 1991 of the Libraries Program of the European Commission.
The objective of the project is a feasibility study of OCR/ICR technology
for retrospective conversion of library catalogs and the development of
specific software for structure recognition in bibliographic descriptions.
The different steps are: the creation of a prototype, the integration of
the prototype in an operational environment and the testing and adaptation
of the model to a real situation. The source catalog is the Belgian National
Bibliography, 1971 which is composed of pre-ISBD records.
The software for structure recognition recognizes the complete structure of 67% of the descriptions. For 27.6% a manual intervention by Jouve personnel is necessary and 4% of the records gave anomalies, which would not conform to the specifications. Those records were returned to the Royal Library for verification.
The output of the conversion are ISO 2709 records in UNIMARC format. The retrospective conversion makes no links to the authority files of the Royal Library.
The project has already ended and the result was quite successful. With a special grant, the Royal Library could order the retrospective conversion of 200,000 bibliographic records. The data came from the Belgian National Bibliography and the Van Hulthem collection. They will be converted by the end of the year; then the UNIMARC data will be transferred to the database afterwards.
3.2. The METRIC Project
At the end of 1993, the Conference of European National
Libraries (CENL) set up Computerized Bibliographic Record Actions (CoBRA)
with funding of the European Commission (DGXIII/E) (http://portico.bl.uk/gabriel/en/cobra.html)
. "CoBRA was to focus on core issues relating to the Commission's library
program and to identify actions needed to promote R & D work which
would stimulate the international services provided by the national agencies
and encourage coherence between national initiatives in order to develop
an EC perspective" (CoBRA, 1995). During 1994 and 1995, five CoBRA initiatives
were funded by the European Commission. One of them is the METRIC project,
which aims at the exploitation of bibliometric data in national bibliographic
databases by creating a common format and a tool for the management of
the national bibliometric data. This tool should enable the libraries to
prepare comparable statistics on their yearly production.
Their are four partners in the METRIC project and four work packages (WP) are specified, so each partner is responsible for one package. The partners and their respective responsibilities are:
• The Royal Library of the Netherlands in The Hague was responsible for the WP2, being the development of the bibliometric application.
• The Deutsche Bibliothek of Germany was responsible for the WP3. The work for both WP2 and WP3 was performed by CAP Volmac. Within these packages the following points were discussed and agreed upon:
- the scope of the data to be delivered, namely data for the years 1994 and 1995
- the description of the format, including the tag, field length, mandatory or not, and the allowed values
- the record format
- the file format
It was decided to use Microsoft Access as the database software for the bibliometric data, so CAP Volmac developed a parser and loading program in order to load data into the database. These database has to be maintained by each of the project partners, and a file server has been set up at the Deutsche Bibliothek where data and software are available via FTP.
• The Swedish Royal Library was responsible for the WP4. In WP4 the project was evaluated. The evaluation pointed out some anomalies in the data: records of years other than 1994 and 1995 (is there a problem with the parser?), some countries have not given, contrary to the agreements, values for certain fields. Some figures should be looked at very carefully: according to the numbers, 7% of the printed publications in Belgium are illustrated, whereas this number amounts to 68% for the Netherlands. Is this a reflection of the reality? There are some other striking divergence among the different countries. Thus, it is necessary for each partner to take statistics for each country, then compare them with the results of the other countries. Reason for striking differences can then be explored.
Although the imperfections are evident, METRIC is
one of the first projects to build a bibliometric database, with different
partners of different countries. The fact that this approach was successful
and lead to a potential valuable work tool should encourage the participating
libraries to continue their efforts to extend their data entry systems
in order to deliver more bibliometric data; it should also attract new
partners for further collaboration.
DELICAT (Data Enhancement of Library Catalogues) is another collaborative project of the Royal Library. During the planning phase the project was named KSYSERROR with an URL address http://www2.echo.lu/libraries/en/projects/ksyserro.html. The project was launched on the May 1, 1996 to promote a greater and more efficient exchange of bibliographic information between the libraries in Europe.
The project aims to create an expert system in order to detect automatically errors in library catalogues. The development and the testing of the system is expected to take two years. The need for this expert system results from the expansion of library computer networks and the increasing tendency to share information resources while avoiding duplicated effort in areas such as cataloguing. However, these shared cataloguing initiatives can lead to the duplication and sharing of defective records, degrading the quality of the local catalogue. Project DELICAT's main objective is to detect those problems by creating an expert system capable of identifying potentially fault data, and warning suppliers and receivers of records.
Full partners in the project are Biblioteca Nacional, Spain; Bibliothèque Royale Albert Ier, Belgium; Fraunhofer Gesellschaft Germany; and The British Library, UK. The coordinating partner is Ifigenia Plus, Spain, and the associate partner is Universidad de Granada, Spain. The Fraunhofer Gesellschaft has experience of integrated expert systems for pattern recognition in industrial environments; Ifigenia Plus has developed expert systems including fuzzy logic and semantic network knowledge representation techniques. The project is funded by the partners and the EC's Telematics Applications Program (DGXIII- E).
This project also is subdivided into four milestones which are themselves subdivided into totally 9 workpackages. WP1, involving a quantitative error study, the error taxonomy and the user requirements is finished already, and WP2, involving knowledge acquisition and system modeling has started.
The quantitative error study is based on the findings of investigations of the British Library, the Bibliotheca Nacional and the Royal Library. Records checked by the Royal Library of Belgium were derived from the database of the Belgian National Bibliography. The following records have been chosen:
- the records for the issue May 1996 (1.081 records).
The chosen records have been proofread and have been checked by computer. The computer program checks on the collation, shelf number, publisher, ISBN, typography such as the presence or absence of capitals or square brackets... Results have been reported in a Quantitative Error Survey (WP task 1.1.)
3.4. EROMM (European Register of Microform Masers) Project (http://www.gwdg.de/~sub/ebene_1/erommeng.htm).
EROMM has been subsidized by the European Commission and was started by the Bibliothèque Nationale de France, the British Library, the Bibliotheca Nacional of Portugal and the Niedersächsische Stats- und Universitätsbibliothek. The EROMM database keeps records of microforms (microfilms or microfiches), in order to avoid the expensive costs of duplication of microforms in libraries. The database can be consulted via the Internet.
Until now, the Royal Library has catalogued nearly 800 microforms, and the data will be sent to the EROMM database, in UNIMARC format, with some extra information relating to microform reproduction.
4. INTERNATIONAL COLLABORATIVE PROJECTS
At the international level, the Royal Library is involved with the International Serials Data System. The number of serials included in the National Bibliography is in the order of 700 a year and these data have to be sent to the ISDS Center in Paris.
The Royal Library, with its many collaborative projects,
has to send data to partners, in different formats. Even if the formats
are standardized, like UNIMARC is, there are still some little, but tricky
and annoying divergence. Software products which make conversions, are
very useful. One of them, USEMARCON, developed by Jouve, Systèmes
d'Information, France seems to be very promising. USEMARCON also is the
result of a European Community project. The product can convert different
MARC formats, as specified by the user. We also look forward to the development
of programs for conversion between MARC format and non-ISO 2709 formats.
REFERENCES
CCB: de collectieve catalogus van België op CD-ROM. (1993). Bibliotheek & Archiefgids, 69 (3): 119-127.
Computerised Bibliographic Record Actions. (1995, March). Fact Sheet 1.
Lupovici, Catherine . (1994). MORE (Marc Optical Recognition). In Library Services in an Electronic Environment: 18th Library Systems Seminar, Budapest, 27- 29 April 1994. European Library Automation Group ELAG, ed. Marta Sipos. Budapest: Nation Szechenyi Library.
Vanderpijpen, Willy. (1994). Het Europees Actieplan voor de bibliotheken en de Koninkijke Bibliotheek Albert I In Informatiebulletin/Koninklijke Bibliotheek, 38 (2): 92-101.
Herpers, A. A. (1995). A Bibliometric Format for
National Libraries in the European Union: PROLIB/COBRA-Metric 10165 Workpackage
1/A. Amsterdam. 75 p.