Exploring Blank Nodes and Local Scopes in RDF Graphs | The Knowledge Graph Conference

The Knowledge Graph Conference Icon

The Knowledge Graph Conference

Dimitris K.
·
RDF 1.1 defines the proposed way for generating IRIs for blank nodes: https://www.w3.org/TR/rdf11-concepts/#section-skolemization
Dimitris K.
·
There are also algorithms for providing consistent IDs (or skolem IRIs) for these blank nodes, see references in https://w3c.github.io/rch-wg-charter/explainer.html
Donny W.
·
they seem to be the only way to correctly apply a "local scope". Is that correct or is there a more appropriate way of applying a local scope?
you could use http IRIs using an authority component that is known to be reserved for local use, e.g. using the .test TLD, e.g. http://mynamespace.test/ as an IRI prefix for RDF nodes
Donny W.
·
it is also a not-uncommon practice to use http IRIs where you control the authority component (e.g. you own the domain name) but you don't actually provide resolution, though they may one day resolve (i.e. they are resolvable)
Donny W.
·
See also https://www.ietf.org/archive/id/draft-chapin-rfc2606bis-00.html#RFC2606 for discussion of reserved top level domain names such as .test and .localhost
Sven V.
·
Thanks Dimitris, I have been looking at skolemization a lot. I do have ambitions of one day publishing my data so I would like to use an IRI that indicates that the blank nodes are to be skolemized later, but also specifies that the blank node identifier is a constant. I guess the second link you shared describes a method of assigning hashes as ids. Is that different from an IRI?
Sven V.
·
Also, the first link https://www.w3.org/TR/rdf11-concepts/#section-skolemization states "Systems may wish to mint Skolem IRIs in such a way that they can recognize the IRIs as having been introduced solely to replace blank nodes. This allows a system to map IRIs back to blank nodes if needed." I guess I am looking for any such established approaches.
Sven V.
·
Thanks Donny, are there any domains that indicate that an identifier needs to be skolemized? What I am looking for is something that indicates to a different entity that the blank node should be skolemized, like a blank node, but is treated as a constant, like an IRI, by reasoners.
Dimitris K.
·
Sven V. these algorithms are used to compute a consistent hash for a dataset but in order to do that, they also compute deterministic IDs for blank nodes (depending on the algorithm you choose). One option could be to use these IDs as a suffix in the skolem namespace you will create, e.g. http://example.com/.well-known/genid/{bnode_id}