MortenHofft
commented
5 months ago Upon request:
How does this work currently?
- search for checklist datasets with the current taxonKey.
- filter that list to only include datasets from a select list of publishers (treatment publishers = plazi) - looking at it now I wonder why we didn't filter upon request for those publishers . e.g. https://api.gbif.org/v1/dataset/search?taxonKey=2440483&type=CHECKLIST&limit=1000&publishingOrg=7ce8aef0-9e92-11dc-8738-b8a03c50a862
- for each of these datasets look for the related species within those datasets using species/123/related?datasetKey=key_from_above_list e.g. https://api.gbif.org/v1/species/2440483/related?datasetKey=58e0c7c5-75ec-4b1b-beb0-d01527f710f9
- if that related species (which is from plazi) has a reference, then it might be a treament e.g.
"references": "https://treatment.plazi.org/id/03A0BE3B6481FF34FF4FFDB4FC4D53CD",- get the species key for that new species key from plazi. get media for that key and the verbatim values. and the dataset and the publisher. e.g. https://api.gbif.org/v1/species/154499077/verbatim
- if the verbatim values have a description that is html and contain a couple of other tells, then we trust that it is a treatment. The main blob of text is the description from the first verbatim extension http://eol.org/schema/media/Document. e.g.
"http://purl.org/dc/terms/description": "<div><p>Orcinus orca (Linnaeus, 1758) ...
All species pages do a search for related treatments. that is currently done in a very ineffienct manner that is also taxing on the APIs.
Currently neither dev, uat or prod APIs are in a state where we can test it. So I'm labelling this as blocked.