Create backend code to generate graph data from RDF

[grtjn]

We need to facilitate generating graph data from RDF. There are example sparql queries in GraphApi.js. It is best packaged in a lib that can be exposed in multiple ways, like rest extension and data service.

The most basic convention used by that code is that any subject/object IRI becomes a node, and any predicate an edge. rdfs:label values are used as label for nodes and edges, with a fallback to a stripped iri name. rdf:type values are used as group for nodes (and edges?), using full IRI if available.

This could be extended with similar conventions for well-known ontologies. For instance SKOS: skos:prefLabel could serve as an alternative for rdfs:label, and skos:Concept IRI as alternative for group. Some vendors, like PoolParty, attach classes to selected topic, a relation that can be used to find ancestor Topics to use for grouping instead of every topic itself. PoolParty uses a proprietary predicated called appliedType for this.

[grtjn]

There is no recommended approach for publishing backend code. Packaging in a jar that works with ml-gradle might make sense. Packaging and publishing as MLPM package could work as well. Or maybe both.