[VEP-003] datalink/core#sibling: Ready for TCG

[Markus Demleitner]

Dear François,

On Wed, Feb 12, 2020 at 06:53:15PM +0100, François Bonnarel wrote:
> Le 12/02/2020 à 14:15, Markus Demleitner a écrit :
> > In December, we discussed about VEP-003
> > (https://volute.g-vo.org/svn/trunk/projects/semantics/veps/VEP-003.txt),
> > which would add a term #sibling to the datalink vocabulary.
> > 
> > Discussion sort of ended with
> > http://mail.ivoa.net/pipermail/dal/2020-January/008257.html -- and
> > now I'm not sure whether there are still doubts if #sibling should go
> > ahead.  Are there?
> Yes I told  you I don't think this covers all the cases where wou want to
> associate a dataset to a source in a catalog.
> Just because as you say below sibling means they have the same origin.
> What we had in mind is something where they are simply cross-corelated.
> 
> Is CrossedDataSet something we could consider a head in branch containing
> sibling, contains, followup ?

Well, "there are additional concepts we could introduce" isn't an
argument against #sibling, which is what we need to make up our mind
about here.  Further terms, including more general ones, of course
can (and should) go in in separate VEPs when use cases for them come
up.

Or do you make the point that the concept "data also derived from
#this's #progenitor" isn't really useful in practice because it's
too specific, too unspecific, or irrelevant to users?  

You see, I somehow have a gut feeling that what we really mean is
"Astronomers who looked at this table row also looked at this and
that other dataset" rather than "this stuff was made from the same
source material", which perhaps is a concept few scientists care
about.

An alternative concept closer to the Amazon use case could perhaps
become a term #see-also.

Anyone out there who has strong preferences for either #sibling or
#see-also?  Perhaps even arguments?  Or an idea on how to decide the
matter in a non-gut-feeling way?  Of course, #see-also could still
become a parent of #sibling later when we pass #sibling now.

So... François, for the very concrete case of the gaia epoch
photometry and the RP/BP spectra -- and of course similar
relationships --, would you object to #sibling and the description
from VEP-003?

Also: Would anyone else?

If not, I think despite my musings on #see-also I'd push VEP-003 to
the TCG.

Thanks,

          Markus