A quick note on exciting work by Andra Waagmeester, Jerven Bolleman, and Hannah Bast. They have created an RDF version of GBIF and made it available via QLever. It is hosted by QLever together with other very large RDF datasets such as Wikidata and Open Street Map. There is an experimental user interface at https://ui.qlever.dev/gbif and if you want to play with the SPARQL endpoint it is https://qlever.dev/api/gbif. The code for generating the RDF is at Micelio/gbif_parquet.
What I like most about this work is that it may help catalyse further exploration of linked data for biodiversity. There have been lots of discussions of the years, and lots of small-scale (typically short-lived) demos, but nothing on the scale of having billions of records available to query. Given how central GBIF is to biodiversity informatics, there’s now an incentive to explore links to other datasets (e.g., Open Street Map, which both Andra and Tiago Lubiana are exploring). It may also motivate people to create RDF datasets for other key sources, such as the Catalogue of Life (CoL) and the Biodiversity Heritage Library (BHL). With this in mind I have created a RDF version of CoL here https://iphylo.org/col/. This endpoint also includes a mapping between legacy GBIF taxon identifiers (integers) and the Catalogue of Life identifiers that GBIF has recently switched too. This means it is possible to have a SPARQL query that takes a CoL taxon identifier and generate a distribution map for that taxon. For example:
PREFIX s: <https://schema.org/>
PREFIX dwciri: <http://rs.tdwg.org/dwc/iri/>
PREFIX wdt: <http://www.wikidata.org/prop/direct/>
SELECT *
WHERE {
VALUES ?col {
<https://www.catalogueoflife.org/data/taxon/D59MS>
}
SERVICE <https://iphylo.org/col/query> {
?gbif s:sameAs ?col .
} ?occurrence dwciri:toTaxon ?gbif .
?occurrence wdt:P625 ?pt .
}`
I have been working on a SPARQL query tool grouse that tries to make SPARQL endpoints a bit more user freiendly, here is that query with the resulting map.
Of course, we can make species distribution maps already, so this merely shows we can replicate existing functionality. The real prize is what sort of questions can we ask that are currently difficult, but would be possible if we had all the core biodiversity data joined up and accessible in one place?
Written with StackEdit.
