DataONEorg / dataone

DataONE information and general-purpose issue tracking
Apache License 2.0
2 stars 0 forks source link

Expose DataONE's holdings as an RDF graph #2

Open amoeba opened 3 years ago

amoeba commented 3 years ago

For our work in GeoLink, we came perilously close to deploying a complete RDF graph of DataONE's holdings. In fact, the graph is still being populated but is kept behind a curtain. This included building a simple crawler with an associated RDF triplestore that encoded the info already contained in resource maps, metadata, and data in an RDF context using the GeoLink ontology. This triplestore allowed us to query our holdings in an alternate fashion but also provided enrichment in terms of co-reference resolution which is a major gap we currently have (there appears to be more unique things in DataONE than there actually are due to a lack of linkages within our federation).

As we annotate the SASAP holdings, we have a new use case for thinking about why and how we might expose semantics within our user interfaces. This could look like running a service that maintains a graph of our holdings and providing some form of query interface based upon which we could improve existing user interfaces or create new ones. To narrow scope, we could consider building something for just the SASAP holdings which could be extended across the federation.

Note: There's active work within DFO to produce a graph about salmon research that's based on Neo4J and LPGs. So thinking about ways to connect this work with an external graph would be useful. Note: We should definitely be looking to integrate with KnowWhereGraph here.