The product-type vocabulary

[Dave Morris]

Hiya,

Why treat events differently to spectra and images ?

On 2021-12-16 13:18, Markus Demleitner wrote:

> .... since I consider a
> difference between identifier and label less desirable than the
> slight awkwardness of having "Event" mean "a collection of [...]
> events", I've already committed it (volute rev. 6063).

> In that vein, "event"
> (that I'm rather sure would in general refer to files containing
> quite a few event*s*) probably should have been called
> event-collection in obscore.

If we follow that pattern then we should register SIAP and SSAP as 
'image-collection' and 'spectrum-collection' ?

A SIAP service is registered with product-type 'image' even though it 
returns a file containing a list of one or more (possibly many) 
image(s).

A SSAP service is registered with product-type 'spectrum' and returns a 
file containing a list of one or more (possibly many) spectrum(s).

If we follow this pattern, a service that returns a list (or stream) of 
one or more (possibly many) event(s) should be registered with 
product-type 'event'.

The vocabularies should be consistent. The easiest (and most portable) 
is to base everything on the singular not plural form.

If you want to keep the plural terms for eventS and measurementS, and 
you also want to keep the singular terms for image, spectra and 
visibility, then to be consistent, add both the singular and plural 
terms for all of them.

     event
     events
     measurement
     measurements
     image
     images
     spectrum
     spectra
     visibility
     visibilities

I do NOT recommend this, but it is consistent.

> .... Having said that, I'd say plural
> nouns in *descriptions* may be a good idea ("a collection of
> events").

To be consistent, should the description of the others also be updated 
to "collection of images" and "collection of spectra" ?

I agree we probably should support plural form in the human interfaces. 
People will want to search for "images", "spectra" and "events", but 
that probably involves a separate stage of pre-processing to convert the 
human form into the standard (singular) vocabulary term.

Cheers,
-- Dave

--------
Dave Morris
Research Software Engineer
Wide Field Astronomy Unit
Institute for Astronomy
University of Edinburgh
--------

> Hi Dave,
> 
> On Thu, Dec 16, 2021 at 12:01:49PM +0000, Dave Morris wrote:
>> The proposed vocabulary mixes singular and plural terms.
>> 
>> The term is event (singular), but the label and description are for 
>> eventS
>> (plural).
> 
> That's easily fixed (it's just a label), and since I consider a
> difference between identifier and label less desirable than the
> slight awkwardness of having "Event" mean "a collection of [...]
> events", I've already committed it (volute rev. 6063).
> 
>> Most of the other terms are singular, cube, image, spectrum, 
>> time-series,
>> visibility etc. but measurementS is plural.
> 
> That's harder to fix, as that's coming down from obscore, and
> changing the identifier would perspectively mean changing what's in
> dataproduct_type.  So... I guess we can't really change this (but
> as I said, I'd like to get rid of the whole concept anyway, as it is
> ill-defined, and it's unclear how it would relate to all the others
> (which are, after all, measurements as well).
> 
>> The example use-case given in the description of the vocabulary uses 
>> the
>> plural term "spectra" in the discovery query, but the target term in 
>> the
>> vocabulary, "spectrum", is singular.
> 
> That language has been a remnant of that earlier experiment, and I've
> removed it as well in rev. 6063.  Having said that, I'd say plural
> nouns in *descriptions* may be a good idea ("a collection of
> events").
> 
>> So a discovery query for a collection of images would search for 
>> 'image' and
>> a query for a collection or stream of events would search for 'event' 
>> as the
>> product type.
> 
> I tend to agree that labels and identifiers in this particular
> vocabulary ought to be chosen so that they're singular; they will be
> used to say what a single thing is, after all.  In that vein, "event"
> (that I'm rather sure would in general refer to files containing
> quite a few event*s*) probably should have been called
> event-collection in obscore.
> 
>         -- Markus