Semantics Calls 5

The fifth edition of our telecon series will took place on Monday, January 17, 2022, at 16:00 UTC.

Here's the mintues:

Introductions

Markus (U Heidelberg), Alberto (ADS), Carlo (Obs Paris), François (CDS), Laurent (Obs Strasbourg), Marco (Obs Trieste), Mireille (U Strasbg), Roberto introduce themselves.

Robert Rovetto (aspiring PhD student, searching for study programs. Independent in meantime. Space knowledge modeling, ontology, conceptual modeling focus) shows some perspectives on how to develop our various semantic resources into a consistent ontology. Carlo mentions that in our UCD development we may have reached a point where becoming more principled might actually help a lot, for instance where meta.ref.doi might now fan out into a lot of individual terms for all the different sorts of PIDs. Making the relationships between them more explicit would likely help a lot.

Baptiste mentions http://ontology.espas-fp7.eu/vocabs; it's not clear to the participants what it is actually used for. He also points to https://earth-planets-space.springeropen.com/articles/10.1186/s40623-016-0542-x.

VEPs"> Current VEPs

• VEP-009 datalink/core#progenitor -- François gives an overview over the what this tries to do, i.e., restricting its extent to non-calibration data. He argues that a recent use at GAVO, where a #progenitor is a datalink document with the raw file as #this and a few #calibration rows would actually perfectly match this definition. Markus is still not convinced it is possible or even desirable to tell apart "calibration" and "science" data, his example being simulations coming with neutrino or gamma data (science? calibration?) – why would a client care whether they are either? François will try to nail down the definitions so this becomes clearer.
• VEP-010 datalink/core#package and VEP-011 datalink/core#thumbnail are uncontroversial.

UCD List 1.5?

Mireille gives a status update: It is on the way, the UCD board has been asked to validate the suggestions.

Carlo asks about PID UCDs (meta.ref.doi siblings/parents) for 1.5. Mireille reports meta.ref.epic, etc are in the review list; how many of these will we need? The use for this would probably that clients give buttons that resolve the IDs, so somehow the resolution strategy needs to be communicated. meta.ref.pid would work for all the PIDs using the http schema – which currently are a lot. ML to ask UCD board advice

Mireille again points out that UCDs help to tag data elements (columns) and it is difficult to do formal reasoning on top of them. She sees them at the leaves of a more structured tree, where UAT manipulates context objects that astronomy is interested in; this would provide a higher level of understanding and reasoning, representing binary stars, HII regions, Pulsar and their behavior, where UCDs just has tags for velocity, position, or temperature.

UCD-related Errata for SSAP and SLAP

The DAL standards SSAP and SLAP require UCDs that are invalid by our rules. Rather than change the rules, we decided to fix the standards through errata. The first already exists:

https://wiki.ivoa.net/twiki/bin/view/IVOA/SSA-1_1-Err-2

Please have a look and review before it goes to the TCG.

arith.diff and arith.ratio in UCDs: Why are they S?

(This was postponed to the next meeting)

The primary part of a UCD is supposed to give a basic idea of what something is. And Markus would say that a ratio of, say, temperatures is very fundamentally different from a temperature – it doesn't even have the same unit. He claims that it has a lot more ratio-ness then temperature-ness. With differences, that's perhaps less pronounced (at least the units match), but still.

UAT adoption note

As part of the vocabulary review, Markus has produced https://ivoa.net/documents/uat-as-upstream/, which has been a PEN for a while now. We should have an internal review of this again and then pass it to the TCG in earnest.

FAIRsFAIR and their criteria for FAIR vocabularies

This was postponed to the next meeting. Volunteers for pushing this ahead would be welcome.

A vocabulary of data product types

François presents a few slides, but the main discussion is also postponed to the next meeting.