Semantics Session at the May 2021 Virtual Interop

Wed May 26, 2021, 5:00 UTC

See InterOpMay2021 for connection infos.

Schedule

Minutes

Product type Vocabulary

See slides. Items discussed:

Dynamical spectra came up at the Radio meeting on Tuesday

Visibilities are hard to categorize; do not really belong to 2d array, also because very specialised clients are required.

Scalar-map doesn't give dynamic spectra a home, it just excludes them; we still need a new terms for them, but of course it'd be great if we knew who would look for them when.

Or course, an array is map from scalars 0..n-1 to a single scalar...

The Planetary Data System has developed a similar vocabulary in the context of their Informational Model to describe data structures. They will post on the semantics list about it.

The sentiment that measurement is not convenient and too large is shared.

Some people cautioned that since this vocabulary should be adopted by obscore, we need to have existing usage there in mind.

The current definition of spectra excludes multi-band spectra or spectra mapping to both flux and polarisation. This is probably not a good idea, in particular because existing spectral clients can deal with this kind of data.

It's not just IRAF that does spectra as 1d arrays, starlink/NDF does that too. (not that there are many of those in the VO, but it's not such a weird way to do it)

Periodograms would be scalar-maps in the current proposal, requiring an extra term (with good reason, as there are extra tools for them).

Concrete example: What would the product type be for these sorts of products? https://archive.eso.org/cms/eso-data/help/1dspectra.html -- since they're consumed by usual spectral clients, we should work hard on keeping them within spectrum.

Status of Object types applied in SIMBAD - Anais

See slides; followup: Sem WG will work with CDS to bring the current state of affairs into the IVOA vocabulary repository.

Vocabularies in Action

See slides.

The extra accept header in the Javascript XMLHTTPRequest is for W3C semantics-style content negotiation: By default, you get back HTML from our repo, to get the convenient json you need to say you accept it (same for RDF/XML or Turtle).