Tuesday, 28 August 2012

The implications of FRBR for e-resource management

FRBR is a framework for organising materials in librarianship and, as the first formal codification of ontological principles in bibliography, it’s a useful tool to apply to a range of bibliographic issues. FRBR (Functional Requirements for Bibliographic Records) was published in 1998 as an attempt to define the relationships between books (and other materials). Basically it presents a description of how the bibliographic universe is arranged (1) and can therefore be used to work out how to organise materials within a library. FRBR says that: 
a work “is realised through [an] expression [which] is embodied in [a] manifestation, [which] is exemplified by [an] item.” 
IFLA, 2009, Study Group on the Functional Requirements for Bibliographic Records

Confusing. One of the best visualisations of FRBR is this one by Thomas ‘orangeaurochs’ Meehan. 

The arrangement of print books is a major preoccupation for librarians: it involves shelving arrangements, classification schemes, day-to-day shelving and straightening, and even architectural design. As E-resources Co-ordinator at Durham University Library, I’m more preoccupied by how to arrange electronic resources and lately I’ve been considering what implications FRBR has for e-resource management. E-resources are ethereal – if they can be said to ‘reside’ at all (2), they do so in a loose network of data connections, servers, and user terminals – but they still need to be arranged and organised and it’s the responsibility of e-resource librarians to consider how best to do this to meet their users’ needs. 

A quick note on defining ‘e-resources’: I’m sort of using the word interchangeably with ‘e-resource records’ since that’s what I actually arrange. We don’t host any ebooks or ejournals on our servers so all I organise are catalogue records and links to external material. By ‘e-resource records’, I mean records for ejournals, ebooks, and databases. 

There are several ways to arrange electronic resources. The catalogue record (or bib record) is the main digital presence of an e-resource: it provides all the data on that resource in the form of a MARC record to AACR2 standard (3). The first way of managing electronic resource links is to add the URL directly into the bib record using the 856 field which, on Durham’s OPAC, then appears like this: 

The advantages of this arrangement are: simplicity; keeping all your data in one place; everything can be done within the LMS’ cataloguing module. 

Another method is to use a separate ERM (Electronic Resource Management) module and to store ERM data separately from the bib records. At Durham, we keep a large coverage database which contains a limited amount of information about every e-resource: which provider they belong to, the title of the item, a unique identifier (ISSN, ISBN, control no.), and – most importantly – a URL. The LMS then creates checkin records which are automatically attached to the appropriate bib records (4). These two records – the bib and the checkin – together form a complete e-resource record and display to the user like this: 

Though more complicated, this method has distinct advantages over the 856 method. Most importantly, the coverage database is the data used by our OpenURL link resolver software which makes resources more discoverable for users. This method also allows us to specify multiple sources for a single e-resource (5): if, for example, the ejournal Analysis is available at more than one website, we simply attach multiple checkin records to a single bib record like so: 

It would be possible to have multiple 856 fields but they wouldn’t display as nicely and, as mentioned, wouldn’t feed into the OpenURL software. 

Other methods for e-resource management are available. If you’re feeling crazy, you can use 856s and coverage together but the result is messy and redundant (6): 

Paul Otlet, the co-creator of the Mundaneum, was a visionary figure in bibliographic thought: he was one of the first thinkers to conceive knowledge as a ‘web’ – a networked knowledge system as opposed to the hierarchical knowledge systems predominant in epistemology and LIS. The foundation of Otlet’s philosophy is that the arrangement of documents mirrors the arrangement of the world. By arranging documents in a certain way, we create a mirror of the abstract world of the ideas expressed in those documents. 

“Everything in the universe, and everything of man, would be registered at a distance as it was produced. In this way a moving image of the world will be established, a true mirror of his memory.” 
Paul Otlet, 1935, Monde, essai d'universalisme: 
connaissance du monde, sentiment du monde, action organisée et plan du monde

In considering electronic resource management, my preoccupation is how best to reflect the world as it actually exists. As well as being an intellectual puzzle, I think asking how to arrange resources in a logical and commonsensical way is the best means of helping users find materials and understand the arrangement of a library catalogue. Catalogue arrangement should reflect the world with which the users are familiar. This requires us to consider the ontology of e-resources and it’s where FRBR comes in handy. 

The first question is ‘what kind of entity is an e-resource?’. Based on the coverage method that Durham uses, an e-resource record is two separate blocks of data knitted together – the bib record and the coverage. FRBR defines an item as a singular entity and so arguably it would be more accurate to keep all the data for an e-resource in one place à la the 856 method. And consider when an e-resource has multiple sources and therefore multiple coverage entries: for example, an ebook that is available on both MyiLibrary and Cambridge Books Online. As copies of the same publication, they seem to be the same manifestation but since they have different digital presences they seem to be different items (7). Therefore they should probably be kept separate but linked in the catalogue: does the method of separate coverage reflect this or should they have completely separate bib records? Looming over all this is the question of what counts as an ‘item’ for an e-resource? 

A larger question is the differentiation between print and digital versions of the same work. One of the issues I’ve been considering is whether, in the long-term, print and electronic records should be merged. The technology allows us to do this and, to some extent, encourages it: rather than keep separate bib records for electronic items, we simply attach the coverage checkins to the existing print bib records. This wouldn’t disrupt the OpenURL software and from a certain perspective it would make resources more easily discoverable to the user: if they’re looking for the journal Analysis, it’s easier to have one record which contains information on the print copies and the electronic versions than two records which must be opened separately. 

Using FRBR, do print and electronic versions of the same work count as separate expressions, manifestations, or items? This partly depends on the state of the electronic version: is it a scan of the print version or a separate copy of the text contained therein? My reading of FRBR suggests that print and electronic versions of the same work are different manifestations which means that the bib records should be kept separate to accurately reflect this. This ontology is fairly woolly and so different interpretations are available. 

This discussion leads to much larger questions about the arrangement of library catalogues. Should we try to the mirror the ontological arrangement of the bibliographic universe? Should we do this according to FRBR’s pre-defined ontology or not? The most important questions are: what arrangement is most useful for the library user and what arrangement best reflects the status of reality? And can there be conflicts between the two? 

(1) More precisely, it presents IFLA’s description of how the bibliographic universe is arranged and may not necessarily reflect the views of others. I have philosophical issues with FRBR but this is not the time to discuss them. 

(2) They can't: that’s a category mistake. But in this kind of ontological discussion, the language gets a little fuzzy and I can’t be bothered to perform the linguistic circumlocutions that would make this post 100% philosophically accurate (2a) so I’m afraid you’ll have to forgive me for any slip-ups I make later on. 

(2a) It would also make the writing dreadfully academic and dull. The philosophers in the audience know what I’m talking about. 

(3) If you're reading this in 2014, replace 'AACR2' with 'RDA'.Or if you're reading this further in the future, replace 'MARC' with 'RDF' or whatever structure has replaced it.

(4) Ideally. Nowadays I’m pretty good at getting them to attach correctly but occasionally e-resource checkins get attached to print bib records or don’t attach to anything which means I have to go through them and attach them manually. Which gets quite laborious. 

(5) There’s an ontological leap in that sentence but bear with me and I’ll get to it. 

(6) By the end of this week, these multiple-URL records won’t display like that in our OPAC anymore. Call me overly cautious but I didn’t want to delete the URL data for 200,000+ ebooks so I’ve been changing the 856 fields into non-displayed custom 956 fields. It’s a bit inelegant but it makes no difference to the user and saves the data in the bib record for a rainy day. The multiple global updates required for this process have been taking a long time and are, as of publication, ongoing. 

(7) I’m assuming that print versions and digital versions are the same expression. My interpretation is that ‘expression as a text in English’ covers both print and digital text.


Jan Erik said...

Regarding your example; "journal Analysis, it’s easier to have one record which contains information on the print copies and the electronic versions than two records which must be opened separately."

You are mixing two different needs into one. You need a datamodel that allow you to describe what you actualy have in front of you. Then you need a good presentation of this model to your users. So you need two records, but you can chose to present them as one to your user.

To many in the library world is lock into the one-to-one mapping of data objects into visual objects. I blame the widespred use of the marc-format, in to many ways...

Mark Perkins said...

If an e-resource has multiple sources, then I would argue that these are multiple expressions NOT different items. When it comes to 'items' in the world of e-resources, these are 'created' at the time of use - very different than the print world.

Working in the French academic system, they have created a field "E856" to enable different institutions to have different URLs for the same e-expression, (proxy logons); the field 856 is then kept a common URL for each manifestation.

Mark Perkins

Simon Barron said...

Thanks Mark. That's a really interesting interpretation: the idea of digital 'items' being created and existing with individual access makes sense and suggests that the ontology of digital objects is more complex than I'd considered.