Everytime I think I've got an angle on this, it seems to slip away from
me...
Alexander wrote:
>
> Amen! Just the fact that trained professional librarians can't agree
> easily on what is what is a *huge* hint to why there is something
> wrong with the current model.
>
In my experience the more trained professional librarians you have, the
more likely disagreement is :)
Anyway, just to throw some more opinion in the pot, for me the FRBR
entities are a reasonably good approximation to levels at which a user
might make an enquiry about something in a library:
Do you have Beethoven's 5th?
Do you have a score of Beethoven's 5th?
Do you have the 1971 Norton minature score of Beethoven's 5th
Can I borrow this item?
This suggests to me that if we were to build systems modelled our data
in this way, it would be easy to build a user focussed interface which
could give answers to these questions.
However, thinking about all this as I was writing this email, it
occurred to me that perhaps I'm coming at this from the wrong direction.
For me FRBR is a useful way of thinking about library stuff, and it
frustrates me that library systems aren't better able to represent stuff
in a FRBR type way very easily. However, perhaps this is the heart of it
- we should be designing systems with data models that would let us
output in a FRBRish way, but we don't need to design systems that model
stuff in a FRBRish way.
Now of course, this is what some products are doing now - but it's
difficult because it is difficult to group the records appropriately. As
I understand it, Martha Yee has argued that the data in the existing
records is good enough to group into FRBRish displays, but others
(notably Karen Coyle) disagree. My own feeling that although some
grouping is possible, the data just isn't consistent enough (and is
never likely to be consistent enough) to make this a particularly good
approach.
I suppose the questions I'm trying to pose are:
Is our primary aim to (a) present information in a FRBR type way, or to
(b) catalogue things in a FRBR type way?
If (a), is (b) a pre-requisite?
If not, are there other approaches that still result in (a) without
doing (b)?
Floating around in my mind is some idea related to James' point about
cataloguing at the item level, and Bernard's point about referencing in
our cataloguing rather than creating static values in our records. Is it
possible to think of a way of creating metadata at in item level, and
still allowing other levels to be formed flexibly?
It's been a long week in FRBR land, and it may be that the above is just
the ravings of someone who has drunk too deep at the well of metadata
modelling.
Owen
Received on Fri Dec 07 2007 - 06:08:55 EST