Hilary Dunlavey wrote:
>
> If the data is stored in US-MARC format and accessible via Z39.50 is
> metadata actually required ?
Great question -- which would be just as meaningful without
mention of Z39.50 -- but recent Z39.50 discussions do help
provide perspective on this. First break down the question,
basically, "if the data is marc, is metadata required?" Is
metadata required for what? For the marc record, or for that
which the marc record describes? One may say "the latter,
of course" but on the other hand, you may want to search for
marc records created on a specific date or catalogued in a
particular language.
Recent Z39.50 discussion of metadata has focused on
searching. The new attribute architecture for Z39.50,
version 1.0, will soon be enhanced by version 1.1 which
will incorporate recent thinking concerning cross domains
searching and metadata. We find it useful to consider the
model:
(a) where a database record *describes* a resource
to be distinct from the model:
(b) where the database record *is* the resource
An example of (a) is a database of marc records describing
books. An example of (b) is a document database.
We make this distinction because most of the thinking
pertains to model (a) but it's important to acknowledge that
(b) is a viable model too.
In model (a), say you have a usmarc record (in English),
describing a French book. Thus the database record has a
language field, whose value is 'French' and that's clearly
metadata pertaining to the *resource*. But what if you want
to search for records in English? There isn't any such
field. (Or maybe there is, but I'm told there isn't.)
So, for this example at least, for the question "... is
metadata required for marc records" -- if we mean metadata
for the *resource* (that is, metadata, beyond what the marc
record provides) the answer is "maybe not". But metadata for
the marc record itself – that's required.
Similary, we distinguish between the date a marc record is
created and the date that the resource is created, the
creator of the marc record versus the creator of the
resource, etc.
In Z39.50 we're developing distinct attribute sets to
incorporate this distinction.
Hilary Dunlavey continues:
>.... (I believe that the Z39.50 spec is
> undergoing expansion to accommodate DC elements ?)
The attribute set/architecture development I referred to
above addresses this, but there isn't any change pending to
the Z39.50 spec.
--
Ray Denenberg
Library of Congress
202-707-5795
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|