Discussion om concept UCD: meta.ref.pid

[Markus Demleitner]

Hi Anne,

On Mon, Nov 08, 2021 at 04:22:36PM -0500, Anne Catherine Raugh wrote:
> The next question I would have is "How is 'PID' defined?". For example,
> every PDS product has a unique URI created by PDS, and if you type it into
> a PDS search you can, theoretically at least, get straight to the
> associated product. But although our archive will persist for generations,
> I still wouldn't consider our PDS product URIs to be PIDs. There's a level
> of broad acceptance and independence from the individual publisher or
> curator that things I think of as PIDs have, that the PDS URIs do not.

First off, I don't think we can fit a definition I'd be happy with
into the very terse definitions we have in the UCD list (these should
normally work in definition-generating machinery, cf. the Explanation
column on
<http://dc.g-vo.org/ucds/ui/ui/form?__nevow_form__=genForm&description=Error&_FORMAT=HTML&submit=Go
for combined UCDs).  Hence, I'd say Yan's "Persistent Identifier" is
about as good as it gets for that purpose.

But for future reference when someone is in doubt, I agree it would
be good to attempt a delineation of what we do and don't mean be PID
in the RFM.  

Now, I can't say I'm happy with that the Wikipedia currently says:

  A persistent identifier (PI or PID) is a long-lasting reference to
  a document, file, web page, or other object.

Mainly, I don't think the "long-lasting" is the decisive criterion
here, as I've personally defined URIs in 1996 that still resolve to
conceptually the same thing (where I'm dodging the nasty question of
identity: How much can change in a document before it changes its
identity?).  I'd say that's long-lasting all right, but these URIs
certainly shouldn't be counted as PIDs.

I'd propose something like "PIDs in general are vendor-neutral
identifiers with a defined curation scheme designed to cope with
reasonable changes in resolution of other identifiers (e.g., DOIs for
documents with different or multiple http URIs) or changes in other
identifying characteristics (e.g., an ORCID for authors with names
evolving due to transliteration or marriage) and a well-defined
resolution mechanism.  Note that normal IVOIDs are *not* persistent
identifiers, as they will change when resources migrate between
authorities." for our rationale.

I'm sure in the context of, e.g., the RDA, much better definitions
have been come up with, which perhaps even address some of the less
hairy problems of "identity".  If someone has a quick reference: I'd
be curious, too.

            -- Markus