Generic FIELD/PARAM metadata items in VOTable

[Markus Demleitner]

Hi Apps,

On Wed, May 17, 2023 at 05:07:59PM +0100, Mark Taylor wrote:
>    (1) Allow FIELD/PARAM to contain INFO children:
>
>         <FIELD name="healpix_id" datatype="int">
>           <INFO name="healpix_order" value="8"/>
>         </FIELD>
>
>    (2) Invent a new element for this purpose, say META:
>
>         <FIELD name="healpix_id" datatype="int">
>           <META key="healpix_order" value="8"/>
>         </FIELD>
>
>    (3) Use the existing LINK element using RDF to indicate semantics:
>
>         <FIELD name="healpix_id" datatype="int">
>           <LINK action="rdf" content-role="#healpix_order" value="8"/>
>         </FIELD>
>
[...]
>
> I think either of us could live with either solution.
> Markus feel free to correct or clarify any of the above.

I think Mark has nicely summarised the state of affairs, and yes, if
someone else does the work I won't stand in the way of either of
these proposals.

Except... I hesitate to complicate the situation, but having somehow
repressed the memories of that discussion, at the Bologna Interop I
actually discussed "proper" RDF (that is, RDFa) in VOTable:

https://wiki.ivoa.net/internal/IVOA/InterOpMay2023Semantics/rdfa-notes.pdf

Given that, I'd suggest that *if* we create a new element as per
META, let's at least make it RDFa-ready *in case* we'd ever want to
go in that direction.  That is: literal values should go into the
element content (if we *really* don't want that because it's not
quite in line with the style of the rest of VOTable, then call the
attribute @content rather than @value), and references would go into
an @href attribute.  And we ought to use @property instead of @key.

So:

  <META property="healpix_order">8</META>

or:

  <META property="healpix_order" content="8"/>

And then perhaps, as an illustration of using an RDF object:

  <META property="https://www.ivoa.net/rdf/voresource/relationship_type#IsDerivedFrom"
    href="ivo://cds.vizier/j/a+a/658/a167"
    >Gobrecht et al: (Al2O3)n, n=1-10, clusters data</META>

Note that that's not enough to produce the right RDF triples (because
the subject will be wrong without doing something on the FIELD, and
that's not pretty to do because of ID vs. id), but at least it won't
build new barriers to RDFa in VOTable.

Given my conclusion in the talk ("it's not low-hanging fruit, and the
fruit's not terribly sweet either"), however, I think I'd still go
for the (totally non-RDFa but otherwise nicely RDF-spirited) option (3).

            -- Markus

(who clearly is still struggling to let go of Semantics:-)