VODML base types : request for enhancement of the IVOA.1.0 template model

[Gerard Lemson]

Hi

-----Original Message-----
From: dm <dm-bounces at ivoa.net> On Behalf Of Markus Demleitner via dm
Sent: Monday, February 16, 2026 5:23 AM
To: dm at ivoa.net
Subject: Re: VODML base types : request for enhancement of the IVOA.1.0 template model


      External Email - Use Caution



Dear DM,

>On Mon, Feb 16, 2026 at 09:39:55AM +0000, Paul Harrison via dm wrote:
>> I think that if a particular model element have a constant UCD then it
>> should conceptually by modelled with the SemanticConcept - but it
>> seems that the type of "topConcept" means that it cannot be used to
>> express the UCD string and there is possibly no conventional value for
>> the vocabularyURI for UCDs. (see
>> https://gith/
>> ub.com%2Fivoa%2Fvo-dml%2Fissues%2F19&data=05%7C02%7Cglemson1%40jhu.edu
>> %7Cadea3d4b87aa455aa66808de6d49ad0a%7C9fa4f438b1e6473b803f86f8aedf0dec
>> %7C0%7C0%7C639068360410214931%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGki
>> OnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ
>> %3D%3D%7C0%7C%7C%7C&sdata=RsdcTEw8EhJAOnaqtPr0m1CfuRKO35Ux04Uh2Td6xDI%
>> 3D&reserved=0)

> First off, VO-DML predates the current vocabulary spec, and therefore limits itself to SKOS.  That limitation should be lifted independently of the UCD business.  I'd be happy to review the language on vocabulary usage for VO-DML 1.1, but I'd need some poking.
> Perhaps we should even have an RDF serialisation for the instances?

>>>
> For the UCD question, however, all that iss not terribly helpful.
> The reason is that UCDs have syntax.  Modelling the UCD atoms as a proper vocabulary is probably possible, and there's even a draft PR for that <https://github.com/ivoa-std/Vocabularies%25>  2Fpull%2F31&data=05%7C02%7Cglemson1%40jhu.edu%7Cadea3d4b87aa455aa66808de6d49ad0a%7C9fa4f438b1e6473b803f86f8aedf0dec%7C0%7C0%7C639068360410244771%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=nocGLqWcEmCZkGlB3hLO7MTFuT2%2BPQzU2hr47o9EUqk%3D&reserved=0>, but you'd want ("compound") UCD words here, and RDF can't (really) do that.

Me, I'm still skeptical that VO-DML should say a lot about UCDs beyond mentioning that Quantity-s can be adorned with them in sufficiently capable serialisations.  But *if* there is a strong reason to say "this must be a valid UCD", that'll need extra logic beyond RDF.
>>>


FWIW VO-DML is not limited to SKOS, see 4.14.1 in the VO-DML spec. In early stages the stereotype was named SKOSConcept which was changed to SemanticConcept. If I remember correctly it was Norman who suggested to generalize it and (as the section is written nicely) he may have written the definition in 4.15. Its meaning is that an attribute with that stereotype should take values from some vocabulary. How precisely it is constrained depends on the kind of vocabulary. topconcept for SKOS, but if one specifies a vocabularyURI it is understood that values MUST come from the identified vocabulary. Could that URI identify the UCD list/vocabulary?
How the value should be expressed is I in my opinion a question of the serialization format. One thing to realize is that VO-DML is not so much about serialization as about annotation. If one annotates some serialization element (say a column in VOTable) as storing an attribute with a <<semanticconcept>> the serialization could mandate that that a URI identifying the concept should be written, but I think for a standard like VOTable one might allow the usual UCDs?




Thanks,

            Markus