Should child events repeat information or inherit from parent

[MortenHofft]

@tucotuco
Similar question as on #122
But this time more broadly and not just locations. What about eventDate, habitat etc.

I ask because I notice that the arctos data is pretty much empty for anything but location and IDs

https://arctos.database.museum/guid/MSB:Host:24585?seid=5159436,https://arctos.database.museum/place.cfm?action=detail&collecting_event_id=11930219,Arctos,https://arctos.database.museum/place.cfm?action=detail&locality_id=11454854,,collection,,,,,,,,,,,,,,,,,collecting_source: wild,,,,,

[tucotuco]

Ideally, every Event should be populated explicitly with the data that apply to that Event. If it is missing, by definition it has to be contained both spatially and temporally within the parent, but there is no further assertion that should be made about it. Of course, if the parent is very specific, the child necessarily has to be at least that specific too, so in these cases it doesn't really matter. I think any property values inherited from a parent must be explicitly flagged as such.

[timrobertson100]

If it is missing, by definition it has to be contained both spatially and temporally within the parent

Is it the dwc:Event definition you mean here, please?
If so, I think we might consider tightening up the language around that (and parentEventID) to be explicit that any child events must be contained within any geography and/or duration provided in the parent Event or any further parents upward.

I'm aware of tracking data discussions where the parent event relates to the location and moment of the deployment of a device, and child events track the route. I don't know if data resulted in that form, but the discussion even happening justifies more clear documentation. I could imagine other groups being similarly creative.

[tucotuco]

@timrobertson100 I was not thinking specifically of dwc:Event, but rather Event as in the Unified Model where it is central to everything. However, I think dwc:Event can serve as the basis for the Unified Model. It has everything needed except eventType, which has passed public review and will probably be ratified soon. If you think we should take the battle to DwC we can.
In the Unified Model, the deployment of the device would not be a parent to the Occurrences the device tracked thereafter. Not unless we change the concept, that is. If we do change the concept, then we'll have to cope with all of the things that "parent" might mean, and we won't be able to "inherit" location or interval information from parents, nor do consistency checks on containment between parents and children.