At 10:30 AM 4/5/00 -0400, you wrote:
>
>I'm surprized at the suggestion that thinking about a successor to MARC
>implies re-writing AACR2.  Since AACR2 is not in any sense an input manual
>for MARC, it mustn't be assumed to be an input manual for any successor
>format either.  AACR informed much of the original logic of the MARC record
>and one can assume that AACR2 will inform much of the logic of the successor
>format.

AACR2 informs more than the logic of MARC -- it defines the semantics of
many of the data elements. If the record is coded as conforming to AACR2,
then you know in great detail what is meant by "title". This is the problem
that Dublin Core is facing; it's relatively easy to come up with a record
format, but it doesn't help much that everyone is using <title></title> if
they're all putting different data between the tags.

MARC has gone beyond AACR2, but the intention has been that it is possible
to encode all AACR2 data elements in the MARC record. Without the
cataloging rules behind it the MARC format would not have been possible to
share MARC records to the extent that we have. The real key is the content,
not the record format.

> MARC records are fiendishly difficult to
>generate from and load into a reasonably well-normalized relational
>database, because of all the exception handling that must be done.

This theme of relational databases comes up frequently in discussions of
MARC and I'm not sure why. There is nothing inherent in XML that will lend
it to a relational model any better than MARC, as far as I can see. The
problem with creating classic relational databases for libraries has more
to do with the data itself than the data transport format. (For more than
you ever wanted to know about library data and relational databases, see
Clifford Lynch's doctoral dissertation:

DISSERTATION
Author:   Lynch, Clifford A.
Title:    Extending relational database management systems for information
retrieval applications / by Clifford Alan Lynch.
Description:    v, 239 leaves : ill. ; 28 cm.
Notes:     Thesis (Ph. D. in Computer Science)--University of California,
Berkeley, Dec. 1987.)

>
>I am therefore in favour of an effort to start the design and specification
>of a successor format to MARC.

I think that a re-design effort would be fascinating and useful. I would
like to see it begin by attempting to define goals (not record structures).
It would need to address the question of semantics and how to facilitate
the great degree to which libraries share data (thus requiring a high
degree of conformity and predictability). Even if we could clearly state
what characteristics we think a next generation of library data standards
should have we would have made great progress. But I maintain that it is
premature to determine the record format before we decide what data we wish
to carry and what we want to do with it.

----------------------------------------------
Karen Coyle                    [log in to unmask]
  University of California Digital Library
  http://www.kcoyle.net        510/987-0567
----------------------------------------------