multiple per-profile graphs (in JSON-LD and Trig)

[VladimirAlexiev]

@griddigit-ci in

need to be able to exchange instance data from different profiles in one instance file and be able to know which attribute is in which part of the data. Maybe using graphs in JSONLD or another trick. Theoretically I should be able to put all data in one JSONLD EQ, SSH, TP, SV and at the receiving end I should be able to only read TP from the JSONLD if I want to do that

• Every CIM triple lives in a Model, which is a named graph.
• So far I've only seen instance files that have exactly 1 model, and from the file naming: that model carries only 1 profile (is that correct)?
• There is no problem to concat several models in 1 file (in CIM XML, Trig or JSONLD), as soon as the standard allows it. These models can carry different profiles
• But do you need to intermix several profiles in 1 model? Then you need to add extra props to capture a "graph-subgraph" relation

While thinking about the relation between models and profiles:

• Does every triple live in 1 profile only?
• I don't think that's the case since we have ontology terms repeated in multiple profiles (up to 19 times if I remember correctly)
• So if you separate profiles by graph, are you ok with having duplicate triples resulting from different quads?
• To give an example: if a resource's rdf:type and MRID are stated in 10 profiles and you put them in 10 named graphs, you'll end up with 10+10 duplicate triples
• This will require careful graph selection in queries, or the need to use DISTINCT in many places, thus making queries more complex and expensive
• You don't want to end up with something like this ERA problem: lots of data duplication (Tunnels) Interoperable-data/ERA-Ontology-3.1.0#117: all triples about the Gotthard-Basistunnel (53km) are repeated 22 times

[Sveino]

Let us not mix up the semantic and the syntax. In general a dataset/graph is represented as one instance file. The exception is DifferenceSet. A dataset can conform to multiple profiles. If we want to exchange multiple dataset we can either use rdfg:Graph or compressed file.

[VladimirAlexiev]

• Vladimir asked: can the base Dataset of a DifferenceModel include data from one or multiple profiles?
• @Sveino answered: the base should always be single Dataset, and it can include data from multiple profiles
• Vladimir: I agree because if you want to make a small change to a complete grid model, that model will include data from multiple profiles.

[VladimirAlexiev]

I've reread this and @griddigit-ci's question was different: can I extract only a single profile from a model?

• Currently we have named-graph per-model
• If you want to delineate statements from different profiles, then we need to use named-graph per-model-per-profile, and a "subgraph" relation (md:DependentOn, dct:requires or rdfg:subGraphOf

In this trig example, <full-model> has no statements of its own (but of course it could have such statements):

@prefix rdfg: <http://www.w3.org/2004/03/trix/rdfg-1/>.

graph <full-model> {
  <full-model> a dcat:Dataset, rdfg:Graph;
    dct:conformsTo <EQ>, <SSH>.
  # md:DependentOn <EQ-model>, <SSH-model> # OLD
  dct:requires <EQ-model>, <SSH-model>. # NEW
}
graph <EQ-model> {
  <EQ-model> a dcat:Dataset, rdfg:Graph;
    dct:conformsTo <EQ>;
    rdfg:subGraphOf <full-model>. # OPTIONAL
  <resource> cim:IdentifiedObject.name "name"
}
graph <SSH-model> {
  <SSH-model> a dcat:Dataset, rdfg:Graph;
    dct:conformsTo <SSH>;
    rdfg:subGraphOf <full-model>. # OPTIONAL
  <resource> cim:SomeSshProperty "1.2"^^xsd:float
}

This can be converted to a single JSON-LD file and will carry the 3 models without any problem.

I'll resolve this and add label spec. If you disagree, reopen with comment