[TopicMapsInLIS] Inaugurating the list

[Alexander Johannesen]

On Liliana Melgar E. <lilimelgar at gmail.com> wrote:
> Maybe TM could be not only a good application for the
> library catalog itself, but to connect all this way to organize
> information within a library...

Not only that, but it could in fact solve the very interop problem our
library systems are riddled with ; it is, after all, a system designed
to handle complex meta data in really smart ways. There's no end to
the things in our world that would fit so beautifully into the Topic
Maps paradigm where the built-in mechanism for merging Topic Maps from
various domains is like a wet dream for any serious librarian. (It was
a wet dream for me, at least, which is why I decided to work for the
library world to see what I could accomplish.)

I've had many stabs at trying to make an ontology for AACR2 / RDA data
(or, if you will, whatever people these days stuff into MARC), but
unfortunately all my efforts break down because of meta data quality
and uniformity, and the complete lack of typed data (which is why it
was so fabulously easy to make a MARC to MARCXML converter ; who needs
types and validation, right? :). The library world knows about the
theories of [persistant] identification, but they truly are terrible
at applying it in practice.

The body of MARC data is fun to search (as in "wouldn't it be fun to
see if we found what we're after" kinda way) but hopeless to rely on
in a typed way. For that to happen all the libraries need to undertake
a "clean up our meta data" international project, and acually agree on
a common set of rules and standards the library world would like to
use. No, I don't mean AARC2, as there are of course various ways,
formally or otherwise, to interpret it to fit some need. I mean, we
need to agree to what it really, really means. DC was on the right
track, and a shame that it only got 15 fields long before given up on.
:)

My best MARC ontology could reasonably well map up / extract data from
oral history records, unless they got beyond 2000 records, which
seriously is too small to be of interest. Some serious work need to be
done in this area, one that I think we *must* do, not for Topic Maps
sake, but for the future of our meta data. We need routines for
creating identifiers across vast materials, for sanitizing field data
(like getting rid of all presentation and notation, or convert
notation to proper fields), rules for merging, rules for FRBRication
of records, rules / routines for breaking a MARC record apart (and how
filtering on those would happen), and on and on.

Of course, this should be ongoing work, not "yet another little
project that looks at Topic Maps." I think it's fair to say that Topic
Maps indeed should be used here, that they are technically proven to
give us huge benefits (especially given the library worlds meta
modeling desires). And not as a a portal of sorts, but as a basis for
dealing with really complex meta data. That's what it's *best* at,
after all.

Anyways, my 0.2$.


Regards,

Alex
-- 
---------------------------------------------------------------------------
 Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------