How do we decide whether to promote a creator to a "source"?

[sarayourfriend]

Some GLAM or research institutions have their own "source" page, like NASA. How do we decide when to create such a source? There are heaps of "sources" that rely on Flickr, like Boston Public Library or even niche ones like New South Wales grassy ecosystems.

How do we decide when to turn a creator into a "source"? Do we have a policy about this? Is it restricted to special-interest societies like NASA or a GLAM institution? Would something like the NSW grassy ecosystems page be a source?

From a search perspective, maybe it's not as important, but from a perspective of making these sources discoverable, I could see that being an interesting angle. For someone wanting to contribute to the commons, maybe if they've got pictures of NSW grassy ecosystems they'd like to know they can contribute to a place that will help them describe the images with accurate metadata?

Just looking for ideas of how we can think about this in the future. Places like Boston Public Library (as an example, other GLAM institutions too) fall in line with our existing patterns, but how do we actually make the decision to create these as "sources"? Is it possible for us to do so without needing to backport the data into the catalogue? What are the core effective differences between a "source" and a "highlighted creator"? Is there any difference, really? Is it possible for a source to have multiple creators? Maybe I'm assuming that there is a 1:1 relationship between a creator of a provider and what we call a "source".

[obulat]

I don't think we have any general guidelines for adding "sources". We do have a DAG¹ that checks all Flickr Commons participating institutions, and suggest adding ones that we don't already use as "source" (or "sub_provider" in the catalog) with at least 300 items with CC0, PDM or CC licenses. We also had a check for Europeana "sub_providers" that was retired.

For context, here is a list of all "sub_providers" in the catalog:
https://github.com/WordPress/openverse/blob/6de944472d188e351af5058675af42f66f12256e/catalog/dags/common/loader/provider_details.py

Just looking for ideas of how we can think about this in the future. Places like Boston Public Library (as an example, other GLAM institutions too) fall in line with our existing patterns, but how do we actually make the decision to create these as "sources"?

Does the current criterion sound good: A notable (GLAM) creator that has more than 300 items can be extracted as a source ?

What are the core effective differences between a "source" and a "highlighted creator"? Is there any difference, really?

Is it possible for a source to have multiple creators? Maybe I'm assuming that there is a 1:1 relationship between a creator of a provider and what we call a "source".

When I first read this discussion, I thought that a "source" is not usually a creator, but an institution. Then I checked Flickr - and it seems that most of the "sources" are also marked as "creators". However, some Smithsonian sources do not have this 1:1 relationship between the creator and a source. Here's one of such examples: https://openverse.org/image/61f31a70-5484-4233-95ab-9298e78ebee8?q=cat (the creator is probably extracted incorrectly, and also doesn't have a creator_url even though it is available on the source site).

Footnotes

1. catalog/dags/maintenance/flickr_audit_sub_provider_workflow.py ↩

[sarayourfriend]

Can we add "check for new proposed sources" to the meetup agenda? Shouldn't take longer than a few minutes, hopefully? The goal is to make it an ongoing practice. Doing it at every meetup is an appropriate interval and the product is just to create issues to add the sources.

[AetherUnbound]

We didn't get a chance for this session during our meetup. What do you think would be involved in it? Going through the list of open issues to look at what sources have been proposed? Or going out to find new sources of CC images to make issues for? If it's the latter, do you have resources or suggestions for how to identify sources?

[sarayourfriend]

Sorry, no, I mean sources not providers. The DAG evaluates potential sources in existing providers, right? The task would just be to go through the DAG's suggestions and flag ones to create issues for. As I understand it, the implementation of a source is just configuration in the area that Olga linked before? Maybe there's more, but it's not the same as a provider..

[AetherUnbound]

Oh, apologies! I think I might be confused because, in the Flickr DAG we have, the DAG itself will alert if there are new sub-providers which have been identified, but only for Flickr. It sounds like you might be suggesting going through our other providers as well for potential sources too, is that right?

[sarayourfriend]

If there are relevant providers then yes, but the key thing I am suggesting is to action on the DAGs alerts at least once a year, regardless of the underlying providers it evaluates.

[obulat]

While reviewing a PR for Europeana script, I realized that we need to promote many of the institutions there to sources. Europeana is an "integrator" that collects data from many GLAM institutions all over Europe, and many of them do fit the criteria above (more than 300 items with an open license or marked with CC0 or PDM).
However, if we do promote all of the Europeana sources, the image filters on the frontend will become unusably long. I wonder if we should have a separate project for outlining the source criteria and automating source extraction and updates, especially from the providers like Europeana, Smithsonian, Flickr or Wikimedia; and improving the way providers and sources are displayed on the frontend.

[sarayourfriend]

I would bet that #753 would be a decent remedy to that by showing the source groups/categories/types above the sources list. I'm not sure if this would complicate things more than necessary, but we could also nest the sources under those groupings as a nested check-box tree (ex. (but not a suggested dependency!) https://github.com/riophae/vue-treeselect)

That might also be helpful if we wanted to retain the Smithsonian sources, it would just create a three-level deep classification (e.g., GLAM / Smithsonian / Smithsonian: Natural History).