I don't think we should use a URL as an identifier. Not in the strict
sense of the word that includes elements like ISBN. You can actually
consider the entire metadata record an identifer of sorts, using
"identifier" in a broad sense. And you really don't know how systems or
people are going to make use of the whole record or individual elements.
So I think that a URL is a location (and a particular kind of location
-- an electronic location) although it may be used as an essential
element in determining what document the metadata describes. The same is
true of publisher, title, date, etc. yet we don't call them identifiers.
The difference is that location is not generally considered part of the
bibliographic description, while those other elements are. But in the
networked world, we sometimes have to stretch to find elements to
describe an item, since we don't have cover pages. The URL may contain
hints relating to the online publication of an item, such as the archive
in which it resides, and that may be as close as we can get to
describing what we would normally think of as a "publisher." (I had
someone argue that a preprint archive is equivalent to a journal title,
and the items in the preprint archive are articles within that journal
-- so lots of different interpretations are possible at this moment in
There is the philosophical question of combining the electronic location
on a record that also refers to a physical location. But that is a
cataloging rule question, and shouldn't alter the metadata format.
On Mon, 2003-09-15 at 06:50, Ray Denenberg, Library of Congress wrote:
> From: "Rebecca S. Guenther" <[log in to unmask]>
> > So, in these situations where the URL is essentially the only reliable
> > identifier, would you suggest putting the data in both <identifier> and
> > <location>? Or just figure that an application would use location if there
> > is no other identifier?
> (I realize you're asking Karen this question, but just to be clear about
> what I'm proposing--) It should go in both.