Reconciliation APIs from other sources

[ddeboer]

Testbench currently reads Reconciliation API URLs from Wikidata. Would it make sense to add other sources as well (for example from a GraphQL API), or is the expectation that they should be added to Wikipedia?

[wetneb]

The endpoints can be added to Wikidata even if they (or their subject database) are not mentioned at all in Wikipedia, so hopefully the hurdle to add endpoints should be minimal (but if you want to add thousands of them it's probably worth thinking if that's an okay thing to import in Wikidata).

The endpoints could absolutely be fetched from another source. Do you have any suggestion of any other database that we should query?

I am not sure to what extent the protocol used is relevant. If endpoints where fetched from a GraphQL endpoint instead of a SPARQL one, in which way would that make your life easier?

[ddeboer]

I mentioned GraphQL because that is what we are currently using for exposing our reconciliation service API URLs. Note that these reconciliation sources are only relevant in the context of the Dutch Digital Heritage Network.

Does it make sense to query our GraphQL endpoint for the testbench?

[wetneb]

Thanks for the clarification :)

Does it make sense to query our GraphQL endpoint for the testbench?

If it would be useful to you, why not! It would probably require adapting the UI to explain where each endpoint is coming from, and maybe avoid showing an endpoint twice if it is present in both sources… so I guess it would be a bit of work to do it well, but as far as I am concerned I wouldn't oppose going in that direction.

Note that this application really is just a test bench. The list of services that it shows is primarily meant to evaluate the adoption of different protocol versions and features, for the purposes of our work on the protocol. OpenRefine started pointing users to it, as a catalogue of reconciliation services, but it's not really suited for this purpose. So if your motivation is to make your services more discoverable to users, it's maybe not a great fit. In my opinion, the better solution would be to develop a user-oriented catalogue of reconciliation services.