VEP-016 date_role#ExportRequested

[Markus Demleitner]

Hi Gilles,

On Tue, Mar 05, 2024 at 03:35:45PM +0100, gilles landais via registry wrote:
> So now for details (see also discussion in datacp mailing list).
> Metadata in DOI and in VORegistry can evolve  - for instance, the landing
> page URL (in bibvo examples the JSON output don't use DOI URL).  It sounds
> obvious that the bibliographic service needs to be aware of changes. So, if
> a new date is needed(?), it should be an "update" date such as the date
> @updated attribute in the registry.

That's actually a good point as to the wisdom of using a date in
VOResource for the export request -- it's relatively straightforward
to request a re-harvest in this way.

This *might* be an opportunity to briefly think about the extent of
changes that are admissible in such a metadata update -- as in: are you
allowed to change relationships?  [I'd definitely say yes]  The
description? [probably yes] The title?  [IMHO perhaps] The authors?
[hm...]  I'd expect there are already guidelines from Crossref or
Datacite on that -- does anyone have a pointer, perhaps?

> More generally, BibVO proposes a JSON linking service independent of DOI
> metadata - To provide this service looks simple to implement, but it asks
> the data center to ensure that metadata in DOI and in the registry are
> consistent.

Umm, that's orthogonal to the whole ExportRequested matter.  The JSON
part
(https://ivoa.net/documents/Notes/BibVO/20240214/NOTE-BibVO-1.0-20240214.html#tth_sEc3)
is about data*sets*, whereas ExportRequested talks about data
*collections*.  Let's keep the discussion of the two mechanisms
apart.

> Date of update for instance, but also relationships existing in DOI
> metadata and in the JSON service. Are there some advises on this point ?

I'm not sure what other communities using DOIs do; there's also the
elephant in the room as to whether datasets should have (first class)
DOIs in the first place.

As things are, Sect. 3 of BibVO is just a slight formalisation of a
mechanism the ADS has been using for a long time to harvest
observatory bibliographies.  It's certainly a good idea to think
about whether there are generalisations that may work within the
DataCite metadata schema, but I think that's a longer-term endeavour
that we would have to pursue in a wider (i.e., DataCite) context.

But as I said: That's orthogonal to ExportRequested, and on a
different level of the dataset/data collection hierarchy.  Let's take
that discussion to a different (datacp) thread.

Thanks,

            Markus