UAT keyword

[gilleslandais]

That's a detail

I propose to encourage dataset provider to use UAT keywords

Lot's of advantages using UAT: which are also encouraged in the registry. It is based on a semantic (UAT thesaurus)

Datacite propose to specify the schema URI whoich can be completed with a classificationCode.

eg:

<subject schemeURI="https://astrothesaurus.org/uat/" subjectScheme="UAT" classificationCode="1054">Milky Way Galaxy</subject>

Question:

• granularity level: to put or not the classificationCode ? (is the UAT url persistent ? can we use classificationCode ?)

[gilleslandais]

After discussion with @BaptisteCecconi, it sounds preferable to priviledge valueURI than schemeURI+classificationCode.

The most often a RDF term has his own URL - classificationCode is an alternative when RDF terms are defined in a same document.(furthermore schemeURI exists already in datacite 4.0)

eg: <subject valueURI="https://astrothesaurus.org/uat/1054" subjectScheme="UAT">Milky Way Galaxy</subject>

[augustfly]

I agree with encouraging UAT keywords. I'm not clear on how to write that advisory when the UAT coverage is variable, esp in helio and planetary @aaccomazzi

[aaccomazzi]

It should be advice that simply says the UAT is the preferred KOS to be used whenever possible. Coverage will improve and the SBN already uses it, so in time I expect more planetary concepts to make their way in (@acraugh please confirm).

A data curator obviously can still add random keywords if they choose, but this will make their resource less interoperable.

[gilleslandais]

a detail: there is here a recommendation a little different in VO registry which proposed to use IVOA-UAT(https://www.ivoa.net/documents/uat-as-upstream/). (SKOS semantics is used to link IVOA to UAT)

however, I don't know if the IVOA semantic added more planetary concept..

[BaptisteCecconi]

The IVOA-flavored UAT is really the UAT, with different URIs. The IVOA-UAT follows the Vocabulary in the VO document, stating that terms (hence URIs) should be human readable. This can be discussed here if you disagree.

We have a script, which takes the UAT as input and produces IVOA-UAT automatically. So the primary source is the UAT.

[acraugh]

Value URIs are intended to be programmatically readable - there is no requirement that they be human-readable, and often they are not (because numbers can be more efficient to deal with than random strings). With respect to the DataCite "classificationCode", this attribute was added as an alternative to "valueURI" for vocabularies that have codes defined in documentation, but do not have APIs to databases that would allow those codes to be programmatically recognized and compared - and thus are not "value URIs" in a practical sense.