Publish first production dataset

[peterdesmet]

Publish first production dataset

[peterdesmet]

First production dataset published at https://www.gbif.org/dataset/20bbd36e-d1a1-4169-8663-59feaa2641c0

[peterdesmet]

The way we handle Events does not work very well with this. We do not model Events. So to show Events we look at occurrences and the distinct EventIDs. And then use the first occurrence within each for the event metadata.

Yeah, the approach worked well when all occurrences (1 HumanObservation, many MachineObservations) shared the same "deployment" event ID. I really liked how it was visualized then.

We gave all occurrences their own event ID to align more with the event centered approach of the Grand Unified Model. I guess the GBIF UI will undergo changes to represent that meaningfully, but can I already suggest dataset pages to show top level parents (only), which the user can drill down in?

For GPS tracking datasets specifically, indicating the number of distinct individuals would be useful. If you want to go advanced, a chart with time (x) and deployment (y) could be useful. Something like:

[peterdesmet]

Note to self: once O_ASSEN is published, reference to it from documentation.

[timrobertson100]

@peterdesmet - would you like us to do anything on how we display this on GBIF for this to go live?

As we designed the data is structured in occurrence core, with a Human Obs record with an eventID for the deployment and then the GPS records captured as Machine Obs each having the parentEventID.

An example of how the deployment event looks like is shown here

CC @MortenHofft @thomasstjerne too

[MortenHofft]

About events in the UI

The way we handle Events does not work very well with this. We do not model Events. So to show Events we look at occurrences and the distinct EventIDs. And then use the first occurrence within each for the event metadata.

So e.g. https://www.gbif-uat.org/dataset/0ef15f32-b41d-4274-ae96-eb5d0059fee6/event/14256071453
Sampling protocol gps comes from the first matching occurrence for that Event.

That also means that Events without occurrences will not appear at all or will have no metadata. E.g.
https://www.gbif-uat.org/dataset/0ef15f32-b41d-4274-ae96-eb5d0059fee6/parentevent/L918401_5742
Since it is a parentEvent without occurrences, then we cannot say much about it (in this specific case we can summarize taxonomy since the occurrences directly mention the EventId as the parentEventId)

We could try to have a fallback and look for the first occurrence with that parentEventId and use that information instead?

UPDATE: above is nonsense, but still true that we do not show the parent event with any meaningful metadata. I've created an issue for it gbif/portal16#1750

[MortenHofft]

Are there anything special that we ought to show for these types of datasets? (ignoring for now how we detect that it is a tracking dataset).

Like the list of distinct organismIDs or something else?

[sarahcd]

fantastic job!

[peterdesmet]

List of datasets is growing: https://www.gbif.org/dataset/search?q=movebank&publishing_org=1cd669d0-80ea-11de-a9d0-f1765f95f18b

O_ASSEN is referenced in the vignette.

Closing issue