species suggest

[MortenHofft]

prompted by https://github.com/gbif/hp-land/issues/5

We have occasionally been asked to provide occurrence search by vernacular names.

I recently added an option to do so using species/search as a suggest. But 2 things could help to improve the results. It might or might not be simple to add:

• filter vernacular name matches by language (so that a search for q=polar bear&language=deu&qField=VERNACULAR would not return anything)
• results for partial matches on scientific name and vernacular names. E.g. polar b returns Ursus maritimus Phipps, 1774 and not empty.

[mdoering]

adding a language filter is rather simple, but partial matches are basically a new suggest search requiring different indices

[MortenHofft]

I suspected. Perhaps that is a better option. A suggest that supports vernacular names. And an option to prefer accepted. Let me try to specify it and come back. Then we can evaluate if it is worth the effort.

[MortenHofft]

q: animalia => returns doubtful genus first accepted names should be prioritized perhaps? https://api.gbif.org/v1/species/suggest?datasetKey=d7dddbf4-2cf0-4f39-9b2a-bb099caae36c&q=Animalia

q: fungi => the fungi kingdom is returned as result 41 Perfect matches should go first I would think https://api.gbif.org/v1/species/suggest?datasetKey=d7dddbf4-2cf0-4f39-9b2a-bb099caae36c&q=fungi&limit=41

q: sea horse => vernacular names aren't searchable Ideally this was attempted, even if it is sometimes incorrect, we could reference the source. And it could have lower priority than the scientific name https://api.gbif.org/v1/species/suggest?datasetKey=d7dddbf4-2cf0-4f39-9b2a-bb099caae36c&q=sea%20horse

q: Ginkgophyta => it is a synonym, but isn't telling me what the accepted name is I would like an option to get the acceptedKey and name and classification back and the synonym name a preferAccepted=true param or such https://api.gbif.org/v1/species/suggest?datasetKey=d7dddbf4-2cf0-4f39-9b2a-bb099caae36c&q=Ginkgophyta

I've made am experimental suggest service in graphql, but it is slow as it have to do a ton of requests to get accepted names, search vernacular names, filter them on language, etc.

It can be tried here https://graphql.gbif-staging.org/graphql?query=%0A%20%20query%7B%0A%20%20%20%20taxonSuggestions%28%20q%3A%20%22Ginkgophyta%22%2C%20language%3A%20eng%29%20%7B%0A%20%20%20%20%20%20key%0A%20%20%20%20%20%20scientificName%0A%20%20%20%20%20%20vernacularName%0A%20%20%20%20%20%20taxonomicStatus%0A%20%20%20%20%20%20acceptedNameOf%0A%20%20%20%20%20%20classification%20%7B%0A%20%20%20%20%20%20%20%20name%0A%20%20%20%20%20%20%7D%0A%20%20%20%20%7D%0A%20%20%7D%20%20%20%20%20%20%20%20%0A%20%20

[mdoering]

All good ideas and great you have analyzed usage patterns. I am just unclear how the integration with CHecklistBank will work and would not want to maintain 2 code bases.

Maybe we can already use the CLB API for some things? https://api.checklistbank.org/dataset/53147/nameusage/suggest?q=Animalia

The usageId values are nub keys, just given as strings cause the identifiers are of any kind across all datasets.

[mdoering]

vernacular name search in CLB has not received much attention yet - but I am happy to change that. There is a separate API for that right now: https://api.checklistbank.org/dataset/53147/vernacular?q=Animal

with taxonID being the nubKey