James Weinheimer wrote:
>
> My question is who should be the "creator". According to your
> definition, Leonardo can no longer be the creator of anything except the
> physical items he produced. If this is so, why is Mark Twain any
> different? With a newly published work of Huck Finn published by
> Cambridge University Press, why don't we put the press as the creator?
>
> The fact is, that's a very strange way of looking at it, and catalogs
> don't work that way. Of course, someone may want Cambridge's version of
> Huck Finn--but that doesn't mean that Cambridge is the creator.
This is where the IFLA model perspective helps:
Mark Twain conceived the abstract work "Huck Finn"
Mark Twain is responsible for the expression of Huck Finn as a Novel
(I think there is a stage version too which was created by someone
else)
CUP published one particular edition of the Novel form of Huck Finn
One copy of the CUP edition of the Novel of Huck Finn is found in your
collection.
Thus, the metadata for the CUP edition should point, through
a Relation, to metadata for the novel, which in turn might point
to metadata for the work. All these metadatas are logically distinct,
but coupled through the Relation pointers. Thus, there is a burden
on the creator of metadata for the derived stuff to ensure that the
metadata for the source stuff exists, or if necessary, create some.
Some systems will allow the metadata for all these _different_ bits
of stuff to be stored in one place - I think the XML syntax for RDF
allows this - see Eric's take on the admin metadata recently posted,
for example.
This raises some significant questions, however:
(1) the DC community must decide how much of the "pure" perspective
of the IFLA model we take on-board (1-to-1 deliberations).
(2) is maintaining all these separate packets of metadata practical
in a _discovery_ context? (clearly "Mark Twain" is a useful piece of
information for discovery for just about any bit of stuff derived
from Huck Finn). In a sytem built according to the IFLA model,
some important metadata might only be visible through following
some pretty extended Relation chains. I worry that the
infrastructure and tools will be lagging years behind the model,
so meantime implementors are likely to continue to kludge away.
--
Best Simon
|