On Nov 30, 2007 6:32 AM, Karen Coyle <kcoyle_at_kcoyle.net> wrote:
> It needs more than a control number. What we really need is a way to
> express the relationships in a clear way.
Yes, indeed. See my other post for more on identification issues; to
create a relationship between things, we must have identification in
some meaningful way. There's way of both rigid, fuzzy and unstructured
ways of achieving identification tokens of various sorts to do this
sort of thing. What I'm keen on doing foremost is to make sure we all
understand the importance of persistent identification, though. I'm
sure you know, but to others it might not be as clear.
> I suspect that there are many
> more relationships than we code in bibliographic records,
It sure is, and there are lots of unspecified relationships as well,
put into notes fields and what have you. The intellectual properties
and semantic richness in our collected MARC store is amazing, but to
do anything automatic on it is really, really hard because one record
speaks of Samuel Langhorne Clemens, another of Samuel Clemens and
another speaks of Mark Twain. These three records have something
*very* important in common which is completely lost on any indexing
search engine. (Notice here that I'm not talking about an author
field, but perhaps some other more obscure field, perhaps in a TOC or
a notes field, or even, if we really want to get our frollicks in a
patter, a full-text field of sorts.)
Everything we do (and I seriously mean *everything* here) is based
around relationships. We take most of them for granted, such as the
link between a cataloger (or the organisation she works for) and the
MARC record she just produced.
Part of the problem with our various automatic systems of the future
is how to create a system of value on our meta data. In the past it
was easy; patron wanted a book, and if library had the book and gave
it to her, success. things are very different now, as goals of both us
and the rest of the world are constantly changing. Even the book as a
transfer of knowledge is dramatically changing. In this ever-changing
environment we need to suss out what we think our infrastructure
(current or future) gives us in value which might be part of solving
our future (and current) issues.
> primarily
> because we have generally relied on humans to supply the semantics of
> the relationship. So if we include an added "author" entry for person x,
> and the statement of responsibility says: "preface by x," then the user
> looks at the record and makes that connection. That doesn't help us do
> anything interesting with that data in a machine environment, however.
Yes, in practice you're absolutely right, although I've hinted to us
creating machine parsing AI to sort and analyze meta data to help
create this added value. We seriously need to clean up the body of
MARC to make the meta data a) more uniform, and b) use more persistent
identification, and c) we all agree to what cataloging standards we
should use (or even a standard for sussing out what cataloging
standard is being used).
One thing that I have learned over the years is that it is easy to
manage a silo of information (i.e. in one ILS), but it's extremely
difficult to do it across several. And *even* harder when going
global. The key to all of these (including the ever popular "federated
anything" services the library world dreams up these days) things sits
squarely inside persistent identification. Let's get crackin'!!
Regards,
Alex
--
---------------------------------------------------------------------------
Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------
Received on Thu Nov 29 2007 - 20:34:33 EST