Subject Re: The other side of academic identity
From Ingrid Melve <ingrid.melve@xxxxxxxxxx>
Date Fri, 25 Jun 2010 10:23:53 +0200

On 25.06.2010 00:56, RL 'Bob' Morgan wrote:
> To paraphrase a famous Netherlands-based CS prof: the nice thing about
> unique identifiers is there are so many to choose from.

> The concept of a unique record/identifier for authors/researchers has
> been around for a long time in the library/publishing community.  I
> don't know the state of the art but it seems there are several competing
> or parallel activities, no doubt with their own histories of politics,
> ownership, regional/national boundaries, etc.  It would be interesting
> to see if the Refeds contacts in the lib/pub space could provide some
> clarity.

We have meet this issue in two different contexts the past few months:
 - the Norwegian national authority registry at the National Library,
implemented by BIBSYS, with support for
 - the research information systems based on
	in our case the new CRISTIN system being developed
In practice we use the National Identity Number (NIN) in this capacity,
but there are a number of shortcomings to this approach (researchers not
all having Norwegian citizenships, not even most Nobel laureates ;)

The topic is on our list of things to investigate, not only because of
the handy unique identifiers, but also because this is a way to track
group membership when people move around.

> If there is a role for Refeds in this, it seems to me that it would be
> in making recommendations to its member federations (and via them to
> institutional IdPs) in handling these identifiers to provide some
> benefit to apps/services.  I would hope this would mean narrowing the
> identifier set to a small number (3 or less perhaps), and defining SAML
> schema for them.  But putting IDs into databases and sending them via
> IdPs is the easy part.

Seems reasonable to share info and keep an eye out in this space.

> The real issue is how these identifiers and their associated records
> (variant name forms etc) are assigned and maintained, and how those
> records are matched to IdM records at institutions.  I'm just guessing,
> but I'll bet this stuff is managed currently by backroom staff at OCLC
> and the like, with relatively poor communications with the identified
> individuals and their institutions, and relatively poor assurance (ie, a
> lot poorer than we do at our campuses).  Could these processes be
> improved by forging strong connections with institutional IdM, using
> federation among other tools?  I'll bet they could.  Perhaps working
> with one or two identifier-sources in this way would help them become
> the preferred ones.

Attribute aggregation, with persistent cross-trackable identifiers used
only where they need to, and real-time one-shot identifiers for
everything else?  I am in.  But we are not quite there yet...