Category Archives: authority control

GND – a new authority file

In May, the German and Austrian library community will make a considerable leap – the new authority file GND (Gemeinsame Normdatei, joint authority file) will come into effect, replacing SWD (subject authorities), PND (name authorities) and GKD (corporate body authorities). So structures which have grown over decades will be supplanted by a more modern approach. SWD, PND and GKD will be merged into one big authority file for subjects, names and corporate bodies, and a lot of historical, from today’s point of view incomprehensible redundancies will be done away with.

So far, there were records for the same entity in at least two of the authority files, and use depended on whether you wanted to control a subject or a name. What additionally complicated matters is that the files had different fields and subfields for essentially the same information. As the German National Library points out, the former formal distinction between authority control for subject cataloging and for descriptive cataloging will be abandoned in favor of a more object-oriented view in the GND.

I see some very good developments in the GND:

  • There will be a set of new rules that align cataloging with international practice and already move in the direction of RDA. A number of details are in line with what RDA will prescribe. Moreover, it harmonizes the rules for authority work that was so far governed by two sets of rules in the German-speaking library community (again one for descriptive and one for subject cataloging).
  • GND will be much more granular than the old authority files. Pieces of information that used to be in a single field are now placed in several subfields. This opens up new possibilities for indexing and searching and for moving towards linked data (the German National Library already has a linked data service in place).
  • Identity management is taken seriously (or more seriously than before, anyway). When you want to say that a person is a musician, for example, you put a link to the authority record for “musician” including ID into the person’s authority record. When you want to say that the Library of Congress is located in Washington, DC, you create a link to that geographic authority record. So all records will be linked with each other and facts are not just expressed by a string of words but by IDs that enable reliable and machine-readable linking. In fact, links are also typed, i.e. you put a certain code for place or profession into the respective subfield to express the type of link that goes out to other authority records.

However, the principle of granularity and linking has not been followed through entirely: we still have something like
100 $P Ludwig $n XIV. $c Frankreich, König
where we could have the information in $c, France and king, split into separate subfields and linked to their respective authority records.

Of course a project like this is a gigantic undertaking which involves a whole lot of work and cooperation. Databases will have to be rebuilt and re-indexed, bibliographic records will have to be updated, discovery tools will have to ingest these new structures, catalogers will have to be trained, and services like VIAF will have to reflect these changes too. Ultimately, though, the GND will make authority work more efficient and easier, and it takes steps into the right direction regarding compatibility with linked data and alignment with international rules.


Authority control and identity management

Kathleen’s comment on my last post made me realize that “identity management” and “authority control” are actually two faces of the same coin. It’s just using two expressions (the first scoped “information technology” and the latter “library”, maybe) for essentially the same thing, although implemented differently. Library identity management has some flaws, the major one probably being that for the most part the identification system has not been brought up to speed and transferred into the digital age.

Moreover, I can think of a number of entities that would be better off with a more stringent IM / AC instead of string matching: serials (which in the German and Austrian cataloging community do have identifiers that are linked to from the records and are treated as kinds of authorities), publisher or place of publication (collocating different names of the same place, e.g. Wien, Vienna, Vienne, Beč, Виена etc.).

Librarians / Catalogers have gathered a lot of expertise in this particular area over the years, we just need to take it a few steps further to fit the online digital environment and build authority control / identity management platforms that can make a real contribution to all kinds of efforts (semantic technologies, digital libraries, digital humanities etc.). One such entity authority tool set that moves beyond traditional library authority practice is EATS, developed and used by the New Zealand Electronic Text Centre. Jamie Norrish anticipated my wanting to bring it up in his comment. 😉 I’ll mention it anyway: see his paper for more in-depth information about EATS (the paper is especially compelling because of its comparison between existing authority control mechanisms and EATS) and another paper he co-authored on the topic maps approach to authority management underlying EATS.

Extending our notion of authority and identity management beyond authors, titles and subjects to other entities in library data creates the opportunity to share links and identifiers with outside communities or across collections and increase search quality through consistency.