On Mon, 21 Apr 2008 11:07:34 +0100, Peter Crowther
<[log in to unmask]> wrote:
> What happens when an institution makes some mass change to its IDs - for
> example when UMIST and University of Manchester merged, and the old
> man.ac.uk and umist.ac.uk addresses all changed to manchester.ac.uk?
> Should there be a similar one-off mass change in Eprints?
That's a policy decision up to the institution, but it would seem
appropriate. The IDs are probably not being exposed as part of the visible
metadata, so it shouldn't matter too much to outside users.
> Similarly, but on a smaller scale, what happens when someone changes
their name (for
> example after marriage)?
Surely that depends on what they want to happen, and on how their wishes
interact with the policy of the repository. They may not be allowed to
retrospectively change their name on already-published material.
> I'm interested, as many people reckon email addresses can be used as
> unique IDs. They definitely can't in the general case, and it's a
dubious
> assumption even in UK academia.
They are probably a bad choice for permanent unique ids. I simplified the
examples I gave in my preceding email. In our eprints.ecs.soton.ac.uk
repository the email addresses are entered by the depositing users, but
they are converted on ingest into a permanent unique staff id. The
permanence and uniqueness are derived from the policies of our Human
Resources department.
By contrast, our institutional repository (eprints.soton.ac.uk) requires a
full staff id to be entered on deposit. Because this is much more difficult
to remember (it is an 8-digit number) there is a separate screen that helps
you identify the member of staff by name search before resolving to the
staff id. Once again, the staff id is guaranteed persistent and unique by
university processes outside the scope of the repository.
---
Les
|