gbif / hosted-portals

Support material for establishing the GBIF Hosted Portals
Apache License 2.0
9 stars 6 forks source link

List of organizations #12

Open tucotuco opened 3 years ago

tucotuco commented 3 years ago

It would be extremely useful for community building to list organizations in a tab in much the same way that datasets are in their tab, with a count at the top of the list.

MortenHofft commented 3 years ago

Is the important part to know which origanisations have published data for a specific filter - that is to answer questions like "who are the publishers that publish datasets containing Flabellina in Mexico"

Or is it more to display the full lost of publishers - possible including publishers who haven't published occurrence data (checklist or metadata only)

dagendresen commented 3 years ago

Somewhere on a portal, the full list of institutions/publishers which publish data in the context of the entire portal is very useful for community building. The list of institutions per search filter would also be useful - but the full list is priority one (on behalf of Living Norway).

tucotuco commented 3 years ago

For VertNet the important thing was to have the full list of participating organizations. No particular need to have the list reflect the current filter in our case, nor would I want there to be confusion on the part of a user seeing a partial list of organization just because a filter was in place.

camiplata commented 3 years ago

We (SiB Colombia) suport this idea, we currently have someting like this https://sibcolombia.net/red-de-socios/ where each organization has some general metrics https://sibcolombia.net/socios/pispesca/, of course for the hosted portal It will have to be different, but just to let you know ;).

MortenHofft commented 3 years ago

So this is interesting - none of you seem to be interested in search/filter on publishers. Just a list/table is what I understand? It is a positive surprise as it makes life simpler.

In VertNets case that would be a list driven from facets (distinct publishers for occurrences). For others it would probably be a fixed filter on publisher search for a country.

No one here seem to be interested in the "who are the publishers that publish datasets containing Flabellina in Mexico" use case.

So essentially this could be a hardcoded list in a markdown table, but as that is hard to maintain, then it would be nicer to drive it from the APIs I gather.

camiplata commented 3 years ago

I do find the filter useful as SiB Colombia has 157 publishers and growing + ~20 per year, I guess for the regional portals the publisher's number will be also high so a long list without filters will be less suited, but the filter can be an upgrade for later.

As other mentioned, for the more 'basic' or most immediate needs of having a list of publishers (https://www.gbif.org/publisher/search?country=CO) and a basic page for each one (https://www.gbif.org/publisher/57dfb71a-f0aa-47c2-b997-30c6048dc98a), I think the API is the way to go to make it sustainable for a long time.

Another option could be to have a more hardwired layout for all portals integrated with the data window such as the one we have for dataset or individual records (I think this is also what @tucotuco had in mind when he opened the issue):

Captura de Pantalla 2021-03-15 a la(s) 2 28 28 p  m

And if some needs something more specific then it can be done with the API/markdown mix.

@MortenHofft you let us know what is posible and wat's not at this first stage

MortenHofft commented 3 years ago

nor would I want there to be confusion on the part of a user seeing a partial list of organization just because a filter was in place.

@camiplata It seems there is disagreement on what is the desired behaviour. I'll have to give it some more thought when the implementation of this starts.

tucotuco commented 3 years ago

So this is interesting - none of you seem to be interested in search/filter on publishers. Just a list/table is what I understand? It is a positive surprise as it makes life simpler.

In VertNets case that would be a list driven from facets (distinct publishers for occurrences). For others it would probably be a fixed filter on publisher search for a country.

No one here seem to be interested in the "who are the publishers that publish datasets containing Flabellina in Mexico" use case.

So essentially this could be a hardcoded list in a markdown table, but as that is hard to maintain, then it would be nicer to drive it from the APIs I gather.

I think there may be three distinct use cases.

The first one is the list of publishers who are participating in the network. This is the one I created the issues for. For VertNet the participants are managed in the GBIF registry at the dataset level. So in theory you could get publishers from the registry in our case, but it wouldn't work for all hosted portals. So your solution to get distinct publishers for all occurrences in the hosted portal (without respect to any other filter) is a better one.

Being able to filter on publisher is a second use case. That could be interesting, but it is not the original issue we posed and is less important in our case.

It seems the third use case is to get the facet for the publishing organization having records in a filtered query. That could be interesting also, but in our case it is less important than the originally posted use case.

MortenHofft commented 3 years ago

@camiplata In your example above I notice that your interest is a list of

Is it correct that you would be happy with the list of Colombian publishers (despite the discrepancy with occurrences) https://www.gbif.org/publisher/search?country=CO and similar for datasets I suppose then? https://www.gbif.org/dataset/search?publishing_country=CO

camiplata commented 3 years ago

Hi @MortenHofft, sorry to arrive sooo late to this thread

MortenHofft commented 3 years ago

Based on this thread I suggest:

Version 1: We start with a simple port of the publisher search functionality on GBIF.org. With the limitations it has. That will meet the needs of @camiplata I believe.

Version 2, will be to support networks that has their own non-searchable publisher endpoint. That is @tucotuco and possible @dagendresen - where publishers are the list of distinct publishers for the constituent datasets for the network. This could include checklists without occurrences.

version 3 will be to find a way to support publisher search for e.g. taxonomic scoped sites. I'm not sure if @dagendresen belong in this or above, but either that would have to be driven from the occurrence index. This will not include checklists without occurrences.