Saturday 21 August 2010

The psychological disorder of the filing system

From the title poem of George Szirtes excellent 2009 collection The Burning of the Books:

Librarian of the universal library, have you explored
The shelves in the stockroom where the snipers are sitting,
The repository of landmines in the parking bay,
The suspicious white powder at the check-out desk,
The mysterious rays bombarding you by the photocopier,
The pyschological disorder of the filing system
That governs the paranoid republic of print
In the wastes of the world?

Hoping that the "universal library" is not a veiled reference to the UL (we hold a collection of Szirtes' letters so I imagine he feels kindly towards us), library "filing systems" are often considered confusing, if not "psychologically disordered", by library users.

The Wikipedia entry tells us that "Classification systems in libraries generally play two roles. Firstly they facilitate subject access by allowing the user to find out what works or documents the library has on a certain subject. Secondly, they provide a known location for the information source to be located (e.g. where it is shelved)".

Perhaps these dual roles are at the heart of the confusion. Unsurprisingly they often prove to be incompatible - subject groupings being countermanded by physical factors like size and space. And for many works, subject classification decisions can seem arbitrary to a library (or indeed bookshop) user. In any case, the majority of bibliographic records contain headings which are able to record subject information with far greater complexity than a single call number.

I have argued elsewhere that users approach our catalogues knowing exactly what they want. If discovery isn't something which normally happens when browsing, then all a classification scheme needs to do is assign a fairly unique identifier to each item and provide a map showing the physical organisation of these identifiers. This is already the case for closed-access collections where browsing is not a factor.

What about classification for online resources, which don't have a physical location? Many libraries either assign a generic call number for electronic material, or don't assign one at all. Discovery is handled mainly through search, and any subject-driven browse facility is generated by the subject headings within the record itself.

Which is all very well in the catalogue, but what if we need to group online resources for other reasons? One of the things we're looking into is pushing subject-relevant content to students. In order to do this we need to "classify" these resources by subject. Some options are:
  • Use existing subject headings in record - BUT these can vary enormously in terms of "width" and "depth" (i.e. how many and how specific they are). They are often assigned by libraries with the nature of their collections in mind - if you only have 10 physics books, "physics" is fine, if you have 10,000 you'll need to be more specific. Can we be sure that subject headings have been applied in a consistent manner across our online resources?
  • Tap into existing forms of recommendation such as reading lists/citations and reflect them in the catalogue - seems the ideal solution BUT difficult to get your hands on the data and link it through to bibliographic records - an "expensive" option.
  • Use circulation data to make recommendations (i.e. "other people on your course borrowed this, other people who borrowed this borrowed that") BUT difficult to track for online material which is not "borrowed" as such - can we easily track access in the same way?
There are wider questions - does this kind of recommendation produce a concentration on core texts at the expense of wider exploration of the collection? Or, handled correctly, might it lead people into areas they would not have otherwise considered?

There are plenty of questions, and I for one don't have any real answers. Perhaps if I hang around by the photocopiers the "mysterious rays" might spark some inspiration ...