[Even list/ event ] stabilizing the name of a data set containing lists of events

[Stéphane Erard]

Dear Mireille, all

> Le 22 juil. 2023 à 21:04, Mireille LOUYS <mireille.louys at unistra.fr> a écrit :
> 
> 
> Dear all , 
> There are currently differences in the way we consider data products containing lists of events in the VO ecosystem .
> This came up in a discussion started about VO high energy data sets last june.


If I remember correctly, the EPNCore definitions were adapted from VOevents at the time of writing. 
A correct way to handle this may be: 
- to reserve "event" to an actual physical event
- to refer to observations of an event in a different way: this is essentially an observation, and several ones may be related to the same event, not necessarily at the same time.

Eg: flyby of Steins by the Rosetta spacecraft. Rosetta observations are dated, says 18:20, while ground based ones are dated 18:40. However, they all refer to the closest approach, which is the event. 
In the solar system, multi-spacecraft observations of events all suffer from this light-travel time lag (think Shoemaker-Levy 9 impact on Jupiter, seen from several locations).
A similar situation may occur for multi-messenger observations. 
And we do have events propagating in the Solar System, reaching various planets after some time. Event = solar event, event_obs = its detection as it arrives at various places.
For completeness: some EPNCore services have dataproduct_type = ev, which stand for event.  

- I think event_list is ambiguous —in my opinion, this refers to different connected events. Think succession of Shoemaker-Levy 9 impacts on Jupiter, from various fragments of a single cometary nucleus: Event = Shoemaker-Levy 9/ Jupiter impact, event list = series of ~ 20 impacts from fragments of SL9.

=> I would reserve "event" for the actual event, and call observations of this event something like event_observation, with an ID potentially connecting them to the event (which allows finding all identified observations of the event easily). 
Therefore, "event" should be reserved to either a prediction, or to the common cause of various event_observations. 
Both predictions (meteor showers) or a first event_observation (supernova…) may trigger a VOevent. 
You could also imagine services listing events and providing related event_observations, from data analysis.

Best wishes
Stéphane


> 
> 1- ObsCore :
> reference document : https://ivoa.net/documents/ObsCore/20170509/index.html
> dataproduct_type = event 
> definition :
> 	• event An event-counting (e.g. X-ray or other high energy) dataset of some sort. Typically this is instrumental data, i.e., "event data". An event dataset is often a complex object containing multiple files or other substructures. An event dataset may contain data with spatial, spectral, and time information for each measured event, although the spectral resolution (energy) is sometimes limited. Event data may be used to produce higher level data products such as images or spectra. 
> 
> comment= considered as a collection of detections with time, space, spectral, polarimetry, etc.  information . 
> should be better named 'eventlist' to highlight that it is multiple .
> 
> 2- IVOA Vocabularies: 
> reference document : https://www.ivoa.net/rdf/product-type/2023-06-26/product-type.html
> definition : 
> 
> event (Preliminary)	Event	A collection of some sort of observed events, such as high-energy particles observed. A row in an event list is typically characterised by spatial, spectral, and time information.	#temporally-resolved-dataset
> 
> comment: this definition also considers a collection of single events. Similar to Obscore event data set. 
> 
> 3- EPNCore :
> reference document : https://ivoa.net/documents/EPNTAP/20220822/REC-EPNTAP-2.0.html#tth_sEc2.1.2
> definition : 
> 	• ev (= event): introduces individual VOevents formatted according to IVOA standard (or possibly events with other formatting). Characteristics are provided via the event_* parameters
> comment: 
>     considered as one individual event , with astronomical context information (ex: FRB,...), propagated in an alert system. 
>     currently the VOevent format is taken into account.
> EPN can gather several event in a catalog of items , each of them beeing an event.
> 
> For clarity,  we need to homogeneize the terms and  definition: 
> so my proposal is to : 
> 
> - update Obscore to clearly use the term  'eventlist' instead of event , as well as the product-type IVOA Vocabulary.
> - change the label 'event' to 'eventlist'  in the product-type vocabulary
> - introduce  'event'  in the IVOA product-type vocabulary as defined in EPN Core.
> 
> 
> Eager to read your comments about this proposal. 
> 
> I'll be on vacation next week and wish you a pleasant summer time.
> Cheers , Mireille
> 
> 
>