At 04:00 PM 4/29/2003 -0700, Roy Tennant wrote:
>Are we visiting the sins of MARC on MODS? That is, are we once again
>building a structure with which we can mimic the layout of a card from
>our card catalog, without thinking critically about what metadata we
>actually need and how to best encode it?

Are we confusing LC's cross-walk and examples with the capabilities of
MODS? There's nothing in MODS itself that seems to require the GMD. And of
course it would never exist in data that doesn't come from a MARC21 record.

Maybe we can make more progress if we parse out this problem into its
1- the MODS structure
2- the MARC-to-MODS crosswalk
3- [other metadata]-to-MODS crosswalks

Bruce is interested in carrying citation information in MODS, so he
shouldn't have to be concerned about where AACR2 (library cataloging)
elements would appear since he won't have them -- unless his citations come
from a library database, but I have the feeling that's not the usual source
for his data. Others of us are indeed taking MARC records and turning them
into MODS, and so we might want to address the crosswalk defined by LC or
develop our own. If, however, Bruce does get data that originated in a
library catalog, he is going to see AACR2 cataloging because that's what's
in those records, just like he may see different citations formats from
different publications. We can get rid of the GMD, but heading choices will
be according to AACR2.

I also think we need to separately address the structure of MODS and the
rules for its content. As you know, this has been the problem area for
Dublin Core, which defines a rather relaxed structure but does not have
cataloging rules that define the content of its fields. MARC is not just a
structure, at least not the way most of us refer to it -- MARC21 implies
the use of AACR2 to determine exactly what goes into a title or author
field. What seems to be happening is that some people are assuming the MODS
also is informed by AACR2, and others have no interest in AACR2 at all. So
we need get clear about this -- will there be a way to know what rules were
used to create the MODS record? Do we know now if the record originated as
a MARC21 record? Is there a way to say that the record was derived from a
PubMed record structure, or an MLA-formatted bibliography.?

I don't think we should go too far down this road, but I fear that if we
don't de-couple the record format and the bibliographic rules we'll
continue to have confusion of purpose.