VEP-015: relationship_type#References

gilles landais gilles.landais at astro.unistra.fr
Fri Mar 22 15:44:25 CET 2024


Thank for the analysis!

In a perfect world, qualifying the relations would be better. In 
practice,  it is not a metadata that we can provide yet (VizieR built a 
database in a  workflow existing from decades).
That's why rdfs:seeAlso (Baptiste proposal) is exactly what we expect 
for VizieR.


The graph you propose (PropX) is a (frequent) relation in VizieR. I told 
about that in a previous mail - however, it is is one among others (and 
not always well flagged in VizieR metadata yet).
Furthermore, this PropX relation sounds too exotic  - I prefer generic 
links !
For me, rdfs:SeeAlso makes the job of GenericRelationship.


Gilles


Le 22/03/2024 à 11:39, Markus Demleitner via registry a écrit :
> Dear Colleagues,
>
> On Thu, Mar 21, 2024 at 05:18:49PM +0100, gilles landais via registry wrote:
>> "rdfs:seeAlso" matches exactly with the VizieR use case -  It sounds to be a
>> good alternative to the VizieR registry "related-to" links.
>>
>> So, please, you can close the  VEP "references" and transform it into VEP
>> "seeAlso" (unless the use of "References" interest other people).
> Ok... I've marked VEP-015 as abandoned; I admit the situation about
> #Cites and #References and all that is too complicated for us to
> unravel, in particular since the beancounters seem to look at the
> concepts, and then the less we have to do with it the better.
>
> However, rdfs:seeAlso is quite certainly not what we want.  True that
> both domain and range of it is RDF resources, and basically
> everything is a resource in that sense and hence *if* we were
> actually doing RDF triples here we could just plug it in and be done
> with it.  But it is so general that it's totally unclear what clients
> should do with it; and it would be *very* hard to delimit its
> extension from what other terms we already have.
>
> As usual, for figuring out sensible meanings it's probably most
> helpful if one first has a clear idea of what a client should do with
> the proposed concept.  What should happen?  Can you draw a mock-up of
> an interface?  How is that interface different from something we can
> already do?
>
> I'm saying this because VEP-016's rationale was *very* far from
> seeAlso (which is basically "everything"); it described a very special scenario:
>
>
>      Paper 1  --------------- (Cites) -------------> Paper 2
>         |                                               |
>         |                                               |
> (isSupplementedBy)                              (isSupplementedBy)
>         |                                               |
>         |                                               |
>         v                                               v
>    VOResource 1 ------------- (PropX) -----------> VOResource 2
>
> PropX in this graph is what we'd want to define.
>
> Is that really the use case you want to express?  Or is it an example
> for "just some old relationship that I don't care to more closely
> define"?
>
> If the former, it sounds what you want would be "Citation by Proxy"
> or perhaps "Supplements Cited of Supplementee" or so.
>
> If the latter, we can simply create a superconcept of all the
> existing terms; perhaps "GenericRelationship".  However, before
> committing so something to general that it's questionable whether any
> useful information is conveyed, I'd really like to see the, if
> you will, user interface to relationships of that sort.
>
> Sorry for being a bit pedantic, but I'm always a bit worried that we
> don't spend enough time thinking about what our consumers should do
> with all the metadata we give them...
>
> Thanks,
>
>              Markus
>


More information about the semantics mailing list