define SHACL Validation API

[VladimirAlexiev]

@HolgerKnublauch in #161

unless we define some sort of validation API similar to the SPARQL protocol.

The spec already

• describes the validation process in prose
• alludes what is passed to a SHACL processor, eg Data Graph

A data graph is one of the inputs to the SHACL processor for validation. SHACL processors treat it as a general RDF graph and makes no assumption about its nature.

I think it's worth specifying a Validation API more formally. It should cater to the following questions:

• Where is the data Need support for RDF datasets (i.e., graphs, quads) #183
• Where are the shapes Need support for RDF datasets (i.e., graphs, quads) #183
• Where is the ontology Where should my ontology go? Data graph versus shapes graph #185
• How many results to generate ability to limit the number of sh:ValidationResults #161
• Nature of results to retain Add severity levels that are not violations #192
• Disposition of results, i.e. where to write them (in memory? new graph; overwrite it? different repo?)
• The test suite currently uses Test Manifest to describe what data to use, what shapes to validate, and what results are expected. (Update the Test Suite #190 asks for additions)
• As part of some work on electrical CIM/CGMES, I'll be describing more complex scenarios (somewhere around https://github.com/Sveino/Inst4CIM-KG/tree/develop/shacl-improved#in-memory-vs-on-disk-databases-and-incremental-validation), eg:
  • A huge electrical Base model (say 100M triples) is validated and stored ("model" is a named graph plus important metadata)
  • A DifferenceModel is being ingested. It refers to the Base model, deletes 100 triples (reverseDifferenceSet) and inserts 200 triples (forwardDifferenceSet)
  • The difference model should be validated during ingestion. This cannot be done standalone, it should:
    • copy the Base to a new graph (no need to validate)
    • apply the differences in a transaction (and validate)

I think that before a Protocol, we should describe an API.
I think web specs use IDL, eg see https://www.w3.org/TR/json-ld11-api/#idl-index .

Assigning to Nick because this task maybe goes beyond the scope of the charter

[VladimirAlexiev]

@nicholascar is this out of scope for 1.2? Then please add some label eg longterm or postponed to it and all/most the mentioned issues.
@HolgerKnublauch suggests to close #161 as out of scope if 1.2. But we need a decision for all/most of these issues, and I'd rather postpone than close them.