Bootstrapping a Biodiversity Knowledge Graph

The "biodiversity knowledge graph" is a nice metaphor for connecting biodiversity data sources, but can we actually build it? Do have sufficient linked available? Given that graph an aggregation of from multiple how do give those sources credit data, and handle changes to data? the classic interface intimidatingly empty SPARQL query box, make within more accessible? This talk discusses attempt with eye on maintain in future. It adopts model similar Global Biodiversity Information Facility (GBIF) CheckListBank where individual providers datasets available as independently citable units Digital Object Identifiers (DOIs). Each dataset comprises form N-triples. To create simply download one or such add them triple store. source assigned its own named graph, provenance each dataset, update any independently. Furthermore, anyone their by mixing matching set (people, publications, taxa, etc.) most appropriate interests. bootstrap this approach, exemplar are created based harvested ORCID, Zenodo, taxonomic name databases. demonstration could be replaced future published directly providers. In some cases there shared identifiers (such DOIs ORCIDs) typically forms isolated islands. help coalesce need "glue" link pairs different identifiers, Life Science (LSIDs) names publications. With addition cross links start generate bibliographies discover communities expertise, more. building also opens opportunities smaller, focussed added using same approach (as N-triples archived online repository). order useful, needs easy visualise. Simply providing endpoint unlikely enough. As part project, I developed GraphQL provide standard queries support simple web graph. provides way explore it being developed, which turn highlight gaps connectivity coverage addressed.