Technical Support hour for Nodes - Global Registry of Scientific Collections (GRSciColl)

Watch the recording here: Global Registry of Scientific Collections (GRSciColl) on Vimeo

In the June issue of the support hour , the Data Products team will give you an overview of the Global Registry of Scientific Collections (GRSciColl). How to edit it in the interface or with he Collections registry API and how the occurrences published on GBIF are linked to GRSciColl entries.

Nodes can either ask a question by replying to this Discourse forum post or write to helpdesk@gbif.org.

2 Likes

Here are some questions that I have for the webinar:

  • How is “Incorporated collections” used? Does the collection to be incorporated needs to be a separate collection in GRSciColl registry?
  • Can user get to the collection information (e.g. contact person) in GRSciColl from Bionomia?
  • Can 1 Occurrence record belongs to multiple collections?

Thank you very much!

1 Like

How does institutionCode and OwnerInstitutionCode are mapped on the GrSciColl registry?

Which fields are mandatory and which ones aren’t?

Best practices for publishing Scientific Collections on GBIF?

2 Likes
  • Are there any best practices regarding Institution/Collection hierarchies? For example: Should a department of an institution be registered as a separate institution or is it better to try and cleanup and have a single institution with several associated institutionCodes?
    -When it comes to collections and collectionCodes I assume that your answer will be that it is up to respective organization to decide on what is collection and how the delimitation should look like? It would be good with some guidelines.

  • Is there a way to connect GRSciColl institutions to ROR - or are there future plans for this functionality?

  • Can you provide some clarity on whether geological collections can or should be registered in GRSciColl, or if it is only intended for biological collections. The name implies that it should be possible.

3 Likes

Is there a way to filter for collections that are not yet digitized in GRSciColl? (like “metadata only” datasets) - i.e. a collection that has no dataset linked to it?

If one collection code is functional in multiple GRSciColl collections,

  • All RBINS Dautzenberg records with a specific collection code: e.g. ‘INV-MOL-Dautzenberg’
  • All other RBINS mollusc records with a generic collection code: e.g. ‘INV-MOL’

could then following two GRSciColl collections be created ?

  • RBINS Dautzenberg collection:
    • main code: INV-MOL-Dautzenberg
    • no alternative code(s)
  • RBINS Mollusc collection:
    • main code: INV-MOL
    • alternative code: INV-MOL-Dautzenberg

Watch the recording here: Global Registry of Scientific Collections (GRSciColl) on Vimeo

Here are the questions and answers that aren’t in the video:

How is “Incorporated collections” used? Does the collection to be incorporated needs to be a separate collection in GRSciColl registry?
The incorporated collections field in the collection entires don’t have to reference other collection entries. You can have one main collection entries with many historical collection cited in the incorporated collections field. Our system doesn’t attempt linking the values entered in that field to other entries.

Can user get to the collection information (e.g. contact person) in GRSciColl from Bionomia?
During the session, the question was rephrased as “can users go from specimens assigned to people in Bionomia to the contact information for the collection holding the specimen?”
Bionomia links to GBIF occurrences, if those occurrences are linked to GRSciColl, then you can access the contact information associated with GRSciColl. I also logged an issue for us to explore how we can improve such links: See here: GRSciColl - explore how to link GRSciColl to Bionomia · Issue #499 · gbif/registry · GitHub

Can 1 Occurrence record belongs to multiple collections?
No, that’s not possible. Right now, a specimen can only belong to one collection.

How does institutionCode and OwnerInstitutionCode are mapped on the GrSciColl registry?
Right now, our system flags occurrences that have a value in the institutionCode field different from the value in the ownerInstitutionCode field. See these examples: Search. In addition to that, the occurrences aren’t linked to GRSciColl because our system doesn’t know where to map the occurrence. Given the feedback provided during the session, we will change this behavior, see : GRSciColl lookup - map occurrence to institutionCode regardless of the ownerInstitutionCode · Issue #496 · gbif/registry · GitHub

In the meanwhile, how to avoid the flag when the institutionCode provided is for an institution and the ownerInstitutionCode provided is for a parent institution (for example a museum within a university)?
In you institution GRSciColl entry, you can have one code as main code and one code as alternative code. This will avoid the flagging.

Which fields are mandatory and which ones aren’t?
Very few fields are mandatory in GRSciColl: name, code and institution. We want to make it easy for people to create entries to fill in later. If we put too many mandatory fields, I am afraid we risk people won’t want to contribute as much.

Best practices for publishing Scientific Collections on GBIF?
We don’t have specific guidelines for specimens publication. What we have is this occurrence data quality requirements page: Data quality requirements: Occurrence datasets. In addition to that, we recommend using collection and institution codes and identifiers that are available in GRSciColl. For more information, please check this blogpost. A general recommendation would be to share as much data as possible. Don’t hesitate to check the Darwin Core Terms available and follow the Data Model work on specimens.

Are there any best practices regarding Institution/Collection hierarchies? For example: Should a department of an institution be registered as a separate institution or is it better to try and cleanup and have a single institution with several associated institutionCodes?
When it comes to collections and collectionCodes I assume that your answer will be that it is up to respective organization to decide on what is collection and how the delimitation should look like? It would be good with some guidelines.
Some answers are provided in the video recording. We don’t have specific guidelines but a recommendation: be practical. If someone is looking for collection information, they likely want to know how to access specimens. With that in mind, it makes sense to group specimens by physical location and/or person who can answer question on how to access them. In the end, it is up to the institution.

Is there a way to connect GRSciColl institutions to ROR - or are there future plans for this functionality?
As mentioned in the video recording, ROR identifiers can be added to GRSciColl entries (there is an identifier type called ROR ). We would like to add ROR identifiers to as many entry as possible and there are ways to do it semi automatically. But we haven’t had the resources to do that so far. That being said, anyone with the proper permissions can add ROR identifiers to institution. You can read more about ROR and GRSciColl and add comments here: ROR and GRID in GBIF/GrSciColl · Issue #294 · gbif/registry · GitHub

Can you provide some clarity on whether geological collections can or should be registered in GRSciColl, or if it is only intended for biological collections. The name implies that it should be possible.
Yes geological collections can be described on GRSciColl. They cannot be linked to occurrence specimens that aren’t fossils but they can be described and added to GRSciColl.

Is there a way to filter for collections that are not yet digitized in GRSciColl? (like “metadata only” datasets) - i.e. a collection that has no dataset linked to it?
Right now, there isn’t really a way to do that. With that in mind, I logged the idea on GitHub, you can follow and comment here: GRSciColl - help node identifier collections with undigitized records · Issue #497 · gbif/registry · GitHub. The closest to a national digitisation monitoring platform we have is the DISSCO UK portal where the number of specimens declared by the institution is displayed next to the number of GBIF occurrences linked to the entry: Data - UK Collections

You mentioned that there is is no hierarchy in GRSciColl does this means that this will come in the future?
We want to make GRSciColl compatible with Latimer Core which models hierarchies. I don’t think we would be able to be as flexible as Latimer Core allows it to be. We might be able to implement some form of structure within collections. But until we start working on it and experimenting with the actual data, we cannot say what will possible and what won’t.

Is there a roadmap document or forum for GRSciColl that we can contribute to?
There is this discourse forum thread as well as the GBIF Registry GitHub repository for more technical discussions and requests (GitHub - gbif/registry: GBIF Registry).

If one collection code is functional in multiple GRSciColl collections,

  • All RBINS Dautzenberg records with a specific collection code: e.g. ‘INV-MOL-Dautzenberg’
  • All other RBINS mollusc records with a generic collection code: e.g. ‘INV-MOL’

could then following two GRSciColl collections be created ?

  • RBINS Dautzenberg collection:
  • main code: INV-MOL-Dautzenberg
  • no alternative code(s)
  • RBINS Mollusc collection:
  • main code: INV-MOL
  • alternative code: INV-MOL-Dautzenberg

Yes, that would work. You would have to rely on collection identifiers to link an occurrence to a collection in that case. Keep in mind that an occurrence can only be linked to one collection.
Note that in your case, since it is about show casing the collection in light of a specific collectors. Perhaps it would make sense to explore ways to link GRSciColl to Bionomia. See here: GRSciColl - explore how to link GRSciColl to Bionomia · Issue #499 · gbif/registry · GitHub

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.