Meeting 2016 03 11

Present

@rubenverborgh, @bergos, @dmitrizagidulin, @niklasl, @elf-pavlik

Content

Dmitri: benefit of calling it "named node" (instead of IRI)

• consistency with “blank node”
  • Ruben: actually, when we say "blank node" in implementations, we actually mean "blank node identifier"
• IRIs have their own spec (overloading the term)
• composition: named nodes contain an IRI, but are not an IRI
• merging nodes based on owl:sameAs and seeAlso (rdflib.js currently implements owl:sameAs, owl:FunctionalProperty, owl:InverseFunctionalProperty)

RDF 1.1 spec refers to them as IRIs

Ruben: who uses "named node"?

• RDF-Ext (RDF-Interfaces)
• rdflib.js
• https://www.w3.org/TR/rdf-interfaces/#named-nodes

Niklas: - what does this API represent? The "bottom level" representation, as in: The graph composed of nodes connected with edges, represented by triples of id:s (IRIs or node IDs or Literals)? - if the node is the vertex (https://en.wikipedia.org/wiki/Vertex_(graph_theory)) it cannot be the edge (predicate)? (That would also be in favour of renaming BlankNode to BlankNodeID I guess) - also consider https://www.w3.org/TR/json-ld/#data-model - considering differences between: Triple(subject=IRI, predicate=IRI, object=BNodeID) Edge(from=Node(label=subjectIRI), label=predicateIRI, to=Node(label=BNodeID)) Resource(id=IRI)r.getObjects(predicate=IRI) == [Resource(id=BNodeID)]

Use cases

Options parameter for the source, sink and store interface

• layers for access control server:
  • ldp http interface -> forwards the identity
  • application -> developer works here
  • uac wrapper store -> uses the identity for triple filtering
  • event wrapper store
  • peristence store

Actions

• wiki page, mailing list, gitter chat, named node vs iri -> dmitri
• example options of the source, sink and store interface -> bergi
• add source sink and store interface proposal to the spec -> bergi