High Energy Physics Libraries Webzine

Editorial Board
Issue 10

 HEP Libraries Webzine
Issue 10 / December 2004

Using FRBR

Knut Hegna (*)



This article presents a possible user interface based on bibliographic data entered according to the FRBR conceptual model.

The main ideas are inspired by the old card catalogue which included a structure in the filing system which was lost in the process of computerization of the catalogues. When pulling out a drawer in the card catalogue you were made aware of the structure by the guide cards, the filing logic and the relations represented by see and see also references.



This article is based on the second half of a lecture on the FRBR conceptual model for bibliographic information given at National Library of Switzerland and at The CERN Library in September 2003. The first part was an introduction to the FRBR Study.

As FRBR is fairly well described in other documents [1], I will not describe it here, apart from saying that the model consists of three components:

of which there are ten, divided in three groups:
  • product entities: works, expressions of the works, manifestations of the expressions, and finally items (copies) of the manifestations. Works and expressions are abstract. Manifestations and items relate to physical publications.
  • responsibility entities: persons and corporate bodies.
  • subject entities: concepts, places, events and objects. The other entities might also appear as subject entities.
the entities are described through the attributes they have. Some examples: a person has a name and birth date; a work has a title; a manifestation has a date of publication and an identification number (e.g. ISBN); an item (copy) has a call number. There are, of course, several more attributes than those mentioned here.
are defined as links or references between the entities. A work is created by a person. "is created by" is a relation between the work and the author of the work, so the author is not an attribute of the work. Some other relations: a work is realized through one or more expressions; an expression is embodied in one or more manifestations; a manifestation is exemplified by one or more items.
There is a lot more to it than described here and I can only recommend reading the full report from the IFLA Study group [1].

Entities, attributes and relations in present catalogues

We start out this journey by looking at a record from the Norwegian national bibliography (parts of the record have been translated into English for the purpose of this article, but still in ISBD):

   Traavik, Morten, 1971-
   Kaptein Marlows testamente : (Mørkets hjerte) / by Morten 
      Traavik ; freely after the novel by Joseph Conrad. - 
      Versjon 3. - Bergen : Den Nationale scene, 1998. - 43
      bl. ; 30 cm. - (839.822[S])
     Theatre manuscript. - "The manuscript is based on the
      Norwegian translation by Sigurd Hoel and the Swedish 
      translation by Margaretha Odelberg" - S.1. - The
      original title of the novel: Heart of darkness. - 
      Performed at Bergen museum, de Naturhistoriske
      samlinger, spring 1998.

Figure 1: Catalogue record for a manuscript by Traavik called 'Kaptein Marlows testamente'.

Putting on the 'FRBR glasses' we see several entities, attributes and relations here.

There are four person entities: Morten Traavik, Joseph Conrad, Sigurd Hoel and Margaretha Odelberg. There are two work entities with the title attributes Kaptein Marlows testamente and Heart of darkness respectively. There are two expressions mentioned: the Norwegian and Swedish translations of Heart of darkness. There is no explicit statement that these translations are of Joseph Conrad's novel, but from the whole setting it becomes clear. The first translation possibly has the Norwegian title Mørkets hjerte.

We see the relation freely after (or "based on") between the two works Kaptein Marlows testamente and Heart of darkness. We might understand from the information given, that "Heart of darkness" is created by Joseph Conrad. We might also understand that Sigurd Hoel and Marghareta Odelberg are responsible for for the Norwegian and Swedish translations of the Conrad novel.

The record and the data are presented in such a way that it can be understood by the human mind and eye. But no computer could ever extract the logical model from these data.

Following some investigations into several library catalogues I was able to make a drawing of the entities and relations shown in figure 2. The boxes are entities (with attribute texts inside) and the arrows are relations.

Figure 2: Works, expressions, manifestations and relations between the two works by Morten Traavik and Joseph Conrad. Boxes are entities, relations are arrows.

During the investigations it became clear that the Norwegian translation by Sigurd Hoel first appeared in 1929 with the title "Det inderste mørke" (Heart of darkness). It was published in a book entitled "Ungdom", another novel by Conrad (Youth). It was republished in 1992 in modernized Norwegian (which makes it a new expression). The 1992 edition was published again in 1999 as a facsimile in another series (which makes it a new manifestation).

The Heart of darkness graph

Having drawn this graph of the two works, I focused on "Heart of darkness" in Norwegian library catalogues. Some of the results are shown in figure 3.

Figure 3: The work "Heart of darkness" and some of its expressions and manifestations, and the persons responsible.

You might add to this structure the film "Apocalypso now!" from Vietnam which is based on "Heart of darkness".

This is the structure, but it would be rather confusing to the reader presenting it in this way, so we may ask ourselves, how could we make the most of it in an understandable way?

Imagine that we performed a truncated search for Conrad among person entities what would the result list look like?

As I am very fond of the card catalogue, I have used this as a basis, adding some graphics and functionality that are now possible with the computer. The idea is shown in figure 4.

Figure 4: The new card catalogue: the result of searching for person=conrad?.

The truncated search for Conrad results in a list of authors (or creators) on the screen along a horizontal axis.

On each 'card' you find the works of the author along a vertical axis.

By clicking on a card you bring it to the front.

The navigational links are highlighted with red-coloured text, so you might focus on either a person or a work.

On the front 'card' for the Joseph Conrad in figure 4 is a list of his novels, but he has also written essays. With this display format you just click on the button on the right-hand side of the card to turn it around to investigate his authorship in full. And if there is not enough room on the back side, we can turn it round once more to find another back side: we can have as many 'back sides' which are necessary to show the person's full authorship.

As this card focuses on the person entity of Joseph Conrad you will also find a link to works about Joseph Conrad, that is, looking at him as a subject entity.

Now what would we expect to happen if we focused on Joseph Conrad by clicking on his name?

We get a list in a similar format, but this time the focus is on the works of Joseph Conrad, one work on each card going along the horizontal axis and the expressions of the work along the vertical axis (see figure 5).

Figure 5: Focusing on works by Joseph Conrad.

You see here that we have links to works about Heart of darkness and links to works based on Heart of darkness. Behind the latter link you'll find the work "Kaptein Marlows testamente".

If the list was a result of a search for title words the horizontal axis would be a mixture of work, expression and manifestation titles. Through the links you would be guided into the FRBR structure.

What the results list should look like must be both a function of what kind of search the user has performed and an analysis of the hits that the search results in. So this interface should be very flexible, but nevertheless lead the user into a structure where navigation is possible.

In the spirit of this interface: if we now choose to focus on the work "Heart of darkness" what would you expect the next screen to be like? The expressions of Heart of darkness are now along the horizontal axis and the manifestations are on the vertical (see figure 6).

Figure 6: Focusing on the work "Heart of darkness".

It is a bit disappointing comparing these wonderful possibilities with what happens in our current catalogues.

In figure 7 you see the default hit list when searching for the Norwegian author Herbjørg Wassmo in a current catalogue. What we see here is an unsorted list. The same work is spread all over, the years are spread, the languages are spread. There is no structure at all, except that all works and expressions are by the same author, which we already knew.

Figure 7: Default hit list when searching for the author "Herbjørg Wassmo".

This is not in accordance with the principle that we should collocate what belongs to the same work entity and differentiate what is brought together.

The system in question can give improved results if we use the option to sort the hit list by title, but there is still confusion amongst results.

Conclusion: Entities and relationships in MARC records

The Traavik/Conrad example above shows that it is difficult (and sometimes impossible) for a computer to identify the entities and relationships in a traditional MARC-based record.

If we look at another record, various entities are easily identified, even by a computer. In figure 8 you find the MARC fields of a record for an esperanto version of Ibsen's play "A doll's house".

Figure 8: "A doll's house" expressed in Esperanto.

Colours indicate attributes of the various entities: the red text concerns the work; the green text concerns the expression, and the blue text concerns the manifestation.

Why couldn't we use this information to present hit lists with a better structure? Eeva Murtomaa, from the Finnish National Library, and I did an investigation into sets of records from the Finnish and Norwegian national bibliographies. The study had a limited scope: fictional literature by single authors. The study showed that, even if there are problems, it was possible to collocate works under the same heading (original title) and expressions under language heading (see the full report [2]). On the basis of structures we found, two suggestions for a user interface were presented in the report. One oh the proposed structures was as an electronic card catalogue.

The idea to collocate records on different entity levels on the basis of different sets of data from the record has since been picked up by Library of Congress and OCLC who now provide free software for FRBRizing library catalogues (see [3] and [4]).


[1] IFLA Study group on the functional requirements of bibliographic records
URL: <http://www.ifla.org/VII/s13/frbr/frbr.pdf>

[2] Knut Hegna and Eeva Murtomaa. Data mining MARC to find: FRBR?
URL: <http://folk.uio.no/knuthe/dok/frbr/datamining.pdf>

[3] Network Development and MARC Standards Office (Library of Congress). FRBR Display Tool Version 1.0.
URL: <http://www.loc.gov/marc/marc-functional-analysis/tool.html>

[4] OCLC Research Activities and IFLA's Functional Requirements for Bibliographic Records.
URL: <http://www.oclc.org/research/projects/frbr/default.htm>

Author Details

Knut Hegna
Informatics library
University of Oslo
Box 1128 Blindern
0316 Oslo

Tel: +47 22 85 25 48
Email: Knut.Hegna at ub.uio.no
URL: http://folk.uio.no/knuthe
Knut Hegna is employed as Senior Academic Librarian at the University of Oslo Library, Informatics department. He has been working with system development for the Norwegian National Library in the 1980's and in the 90's experimenting with user services from the pre-web area and to this day. He has also been lecturing an introductory course on object-oriented programming.

For citation purposes:

Knut Hegna, "Using FRBR". High Energy Physics Libraries Webzine, Issue 10, December 2004.
URL: http://webzine.web.cern.ch/webzine/10/papers/1/

Reader Response

If you have any comments on this article, please contact the  Editorial Board
Editorial Board
Issue 10

Last modified: November 2004