Finding accurate IDs I withdrew due to the way observation labels used to work

[bouteloua]

Is there a way to find the IDs I may have withdrawn in the past due to the old way IDing used to work?

For example:

Before the “Observation Taxon” (what displays at the top of the page) and the “Community Taxon” (what displays on the right side) were split apart, if someone else refined this obs from class to family, I would have withdrawn my coarse ID so that the observation could be labeled as family. If I did not withdraw it, it would still be labeled at class until another person IDed it to family.

I must have hundreds of these types of withdrawn, Improving IDs. Is there any way to find these types of IDs? Or is this the closest I can get? http://api.inaturalist.org/v1/identifications?user_login=bouteloua&current=false&category=improving&order=asc&order_by=created_at

That one will include observations where I withdrew an ID but added another ID, like this: https://www.inaturalist.org/observations/158099 (I don’t want to see those types)

There’s no way to find obs where I withdrew an ID but did not add a new ID, right? thanks!

[aisti]

I’m new to the API, but would it be possible to make a second request for your IDs that have not been withdrawn, then “subtract” the withdrawn-then-added IDs from the first results if they are on the same observation but have an earlier timestamp?

[sessilefielder]

You’re on the right track. I think we want a set difference of { http://api.inaturalist.org/v1/identifications?user_login=bouteloua&current=false&category=improving } - { http://api.inaturalist.org/v1/identifications?user_login=bouteloua&current=true }, i.e. all observations with a withdrawn “improving” ID by bouteloua without a corresponding current ID by bouteloua.

You can also winnow down the first set by flooring the ID rank at a certain level, e.g. https://api.inaturalist.org/v1/identifications?user_login=bouteloua&current=false&lrank=class will only return withdrawn IDs by bouteloua with at class rank or higher.

[sessilefielder]

Actually, it’s even easier. The results for http://api.inaturalist.org/v1/identifications?user_login=bouteloua&current=false&category=improving include all identifications on each observation, so you just have to iterate through all the returned observations and discard any that have an identification by you with current = true.