export to text or Oracle?
Tim Mallory
tmallory at timberland.lib.wa.us
Thu Oct 3 12:48:46 EDT 1996
Fellow Listoids - Karen's post brings together, at least in implication,
several evolving ideas. I see these as the achievement (IMHO :)>) of
another rung on this infinite intellectual ladder we're climbing (god i
hate this pedantic bu******) . . .
1. Metadata is necessary. As appealing as full-text searching is to
technoids, any author's writing is only as clear as their thinking. Hence
full-text searches retrieve immense irrelevance as well as nuggets.
Cataloging is not dead, but it needs to evolve very rapidly right now.
2. Cloning "new records from old" is necessary. The ability to do this
electronically is wreaking havoc with copyright law, and author's
attribution and recompense, but the same technology has enabled "copy
cataloging by clerks". I have to be careful here. I am very resentful of
the idiotic mistakes that are perpetuated this way, with very little
involvement by MLS (read suitably trained to make correct judgements)
professionals, while recognizing that this enables our current diaspora of
reasonable electonic access to almost everything. What is necessary is a
way for cloned records to evolve, to meet the local needs of each
locality/person/priority. We each need a personalized "access-ory" for
our information needs. Ultimately, our personal electronic agent will
know our information wants/needs, but the datauniverse we access needs to
be "customized" for everybody, not just the agent serving each item's
creator.
3. Since we can all change what we access electronically, the metadata
for the items we access can also evolve. Full-text searches key onto
certain terms, which in an ideal datauniverse would be the terms present
in the metadata for fastest access. Can we create documents with the
ability to capture search terms, adding them to the metadata accompanying
the item? This evolving metadata field should have a capability of
prioritizing, moving the most-frequently-asked terms to the beginning of
the metadata.
4. To utilize this evolving metadata, we need mutable search engines.
The first choice would be between "good enough" and "exhaustive". Most
search engines now are set on "exhaustive" so my public-library patron
gets 117,641 hits when trying to find the name of a national politician. A
"good enough" search engine would stop, in a few seconds, after locating a
set number of hits (5? 50? their choice) where the matched term was in top
relevancy tier. This SUGEF (Search Until Good Enough Found) engine would
be a godsend for public libraries, while researchers would still want
exhaustive searches, or use free-text to find "Everything In The Whole
Wide World" (the EITWWW engine).
The Caveats - as if you couldn't tell, I am a public librarian. I
seldom have time for this kind of rumination, as almost all of my time is
spent directly serving the public. My apologies for possibly
inappropriate rants, as I don't have the time for an exhaustive search to
see if what I postulate already exists. I am certainly jealous of several
of you who have the time to contribute regularly to this list, and I
certainly appreciate the evolution of ideas in which I am an avid lurker.
Whether or not I have the skills, I certainly do not have the time to
implement just about anything beyond another Internet apology sign made
with PrintShop. I appreciate the community that creates the tools I use.
----------------------------------------- _.. ... __... _.._____
Tim Mallory, Reference Librarian
Aberdeen Timberland Library Voice: 360-533-2360
121 E. Market St. FAX: 360-532-2953
Aberdeen, WA 98520-5292 e-mail: tmallory at timberland.lib.wa.us
__%\__"round and round goes the wind" - ecclesiastes i:6 __%\__)))))>>>*
On Thu, 3 Oct 1996, KAREN SCHNEIDER wrote:
> I am mulling over a scheme for cataloging Internet resources and
> exporting them to our local web database, soon to be
> Oracle-based. (GILS is the Government Information Locator
> Service--see http://www.usgs.gov/gils/ for more info.) In fact, I
> am mulling over a second scheme for exporting them from GILS
> into OCLC (since many of htese records will be created with a
> metadata entry form on the web by the document originators and
> will then pass to the Library for QA). Well, as long as we're
> indexing and ctaloging, and fiddling and twiddling, and having
> meetings and rewriting stuff, why not, ay? I'm not clear it
> wouldn't SAVE time to create the records in OCLC and export,
> anyway, assuming we got a good workform going.
>
> We're using locally-developed subject terms (which the GILS
> profile supports) , developed by the Region 1 Library, but our
> records, in accordance with the GILS profile, map to USMARC
> (though normal human beings wouldn't realize this).
>
> As I mull this over, I see at least several issues:
>
> * Ensuring that records enttered locally can export cleanly to
> proper OCLC format -- by this I mean, for example, if we are
> entering dates, making sure they are entered IAW USMARC
> format, so we don't have to clean them up. This also, quite
> frankly, has to do with our cataloging expertise, which is minimal;
> frnakly, I've shyed from Intercat because I didn't want to
> contribute incredibly bad records. I'm almost to the point where I
> can clone a new record from an older record.
>
> * Figuring out how to export from text or Oracle to MARC
>
> * Figuring out how to export from OCLC to text or Oracle
>
> What I'm trying to do is comply with the spirit of Intercat *and*
> GILS *and* local indexing efforts. Our libary catalog, quite
> frankly, isn't going to be the place people look for Internet
> resources, nor is it a good place to do this anyway, being a
> primitive mainframe proprietary catalog. (In other words, when
> you get there you can't do anyting about the URL, unless your
> client makes URLs hot, which isn't something we should impose
> on the user.) And this isn't something we control or can change
> locally, either. But I still see it as potentially valuable (even from
> a strictly exploratory perspective) to enter these records into
> OCLC, if possible, and is an interesting example of controlling
> records with multiple destinies.
>
> Any suggestions? I wish I could show you the metadata entry
> form--I'll take some screen caps and put it on our web page. (It's
> not accessible to the public.) It's pretty neat, and was developed
> in a coordinated computer-weenie-library-weenie effort.
>
> Karen G. Schneider
> Contractor, GCI
> Director, US EPA Region 2 Library
> opinions mine alone
>
>
More information about the Web4lib
mailing list