Dear GBIF
I have noticed there’s a certain ambiguity/shortcomings in the behaviour of the GBIF Registry API https://api.gbif.org/v1/dataset/{key} endpoint. I am hoping you can provide some assistance and clarification regarding its intended use and potential alternatives.
The response of https://api.gbif.org/v1/dataset/{key} endpoint contain properties taxonomicCoverage, geographicCoverage and temporalCoverage. In many cases, these seem to be not provided, even if all the occurrence records in a given dataset do contain relevant temporal, geographical and taxonomic descriptions.
Are the contents of taxonomicCoverage, geographicCoverage and temporalCoverage user provided rather than machine-derived? If so, why? If so, is there an alternative way to programmatically (preferably via an API) retrieve these for any occurrence dataset published on GBIF?
Some examples. Both 0f0df35b-e431-4709-8b7a-9496fff9f253 (Anopheles stephensi occurrence data 1985 to 2019) and fe388907-e2b6-44be-a9cb-a80b3924284c (gbif.org/dataset/fe388907-e2b6-44be-a9cb-a80b3924284c) contain 100% records with taxon match, over 90% records with coordinates and over 60% records with temporal information. Yet only 0f0df35b-e431-4709-8b7a-9496fff9f253 provides taxonomicCoverage, geographicCoverage and temporalCoverage and fe388907-e2b6-44be-a9cb-a80b3924284c provides none via the API. Because both datasets do contain this information, it should however be possible to machine derive this for both and make it available via the API for both.
The current design prevents certain programmatic processing use cases using the API.
Similar but not the same also applies to property countryCoverage.
Thank you for your help and insights. 
Stanley
Software Engineer @ Imperial College London