On Tue, 2004-11-30 at 07:31, Bruce D'Arcus wrote:

> > the SAME identifier by different people/databases, and there is no
> > universal identifier that currently is one-to-one with the article.
> Well, DOI.

Far from universal -- there have been DOI's assigned to many articles,
but nothing like the whole universe of articles. Remember that the D in
DOI is "digital" and it is mainly used for items in digital formats.
Many citations will be to articles that are in print formats only.

> And what of isbns or lccns?

The ISBN is essentially a product number for publishers. Every publisher
that has put out a version of Moby Dick has given it a different ISBN.
It's not unique to the work. It's only for books. (Well, and the
occasional teddy bear, but that's another story.)

The LCCN identifies the machine-readable record created by the Library
of Congress. Again, not a work identifier, and there aren't LCCNs for
anything that LoC hasn't cataloged (i.e. the entire universe of

Basically, if you have an ISBN or an LCCN, you should keep it -- it
might be useful for locating the item in another database. But it's
hardly any guarantee that you have a shared identifier with someone


> as you say, this doesn't help for all of those items that don't have
> these sort of identifiers, but my thinking is that info solves one
> obvious problem in that it actually identifies the type of a given
> identifier, and so gives some flexibility to use different kinds in a
> single system.  So, if a DOI is present, that would be used, and the
> fallback might be SICI.
> A tricky problem; this.
> Bruce
Karen Coyle
Digital Library Specialist
Ph: 510-540-7596 Fax: 510-848-3913