In keeping the metadata close to the citation, as with unparsed lists of
authors and part information, wouldn't we lose the ability to transform
the data?  Consider, e.g., a list of four authors + et al. in the XML
record, when the desired output style requires three authors + et al. Or
unparsed part information like "4(3 suppl 2)" when the output style
requires volume numbers, but not issue numbers, to appear in boldface.
The elements will ultimately have to be parsed; although the original
citation could be preserved in <recordOrigin>.

A question with respect to the et al. issue: If all you had were a
secondary source with a partial list of authors, and no access to the
primary source, what would be the best way to encode that in MODS?
Maybe link to a "special" record in the name authority file for "et
al."?  I haven't found myself in that position yet, but it's bound to
happen someday.

--Andy

>>> [log in to unmask] 2004-10-30 11:38:14 >>>
On Fri, 2004-10-29 at 19:03, Bruce D'Arcus wrote:


> If you have any specific thoughts Karen, let me know.

Well, I know that it would be a great deal of work, but I really think
we need a "citation" format -- something that is loose enough to carry
the data from the wide variety of citation styles (and non-styles ;-).
Depending on your need, you could use multiple fields for multiple
authors
  <creator> Jones </creator>
  <creator> Smith </creator>
  <creator> Doe </creator>

Or you could have an author statement like those seen in many
citations:
  <creatorList>Jones; Smith; Doe; et al</creatorList>

At least in this way you know what you have, and you don't have to
parse
the field into separate names if you don't have that need. (Because
parsing always adds a risk of error or loss, i.e. what do you do with
that "et al" in MODS? yet it carries information.)

And a citation format would have similar ways to encode the volume,
number and other designations for serials -- separately, if that's
available, or as a general statement, if that's all that the citation
provides. Again, the point is that you want your data elements to be
"true" to the data in them -- you want to know if the "volume number"
is
just the number or if it has the caption ("v.3"). You want to know if
you have just the first page number (which is all many indexing
services
give you), if you have the first and last, or if you have the number
of
pages (which is included in some services).

I always get nervous about shoe-horning one set of metadata into
another
because the meaning of the data elements and what's in them gets
fuzzy,
and that makes it hard to do elegant programming. You lose information
about what your data looks like and what it represents. I don't feel
that data has to be perfect; I do want the data and the data elements
to
be in sync.
--
-------------------------------------
Karen Coyle
Digital Library Specialist
http://www.kcoyle.net
Ph: 510-540-7596 Fax: 510-848-3913
--------------------------------------