In my opinion modelling the occurrence as a blank node does not improve over the current Schema.org approach and introduces dependence on something which is not yet standardized. How would this work in microdata?

What do you mean with “current schema.org approach”? Not entirely sure what you’re saying here 😅

I mean using the Role class combined with double predicates as a means to reify the relation.

Sorry, still don’t get it 😅

Please refer back to earlier in this thread where you both talk about using EmployeeRole which works nicely for @jdevalk only the double worksFor is weird. I referenced the section from RDF-star draft where they describe the difference between an RDF-star triple and occurrence.

In case of <jdevalk> <worksFor> <yoast> but you want to describe he held different roles at different times, you cannot just say << <jdevalk> <worksFor> <yoast> >> :started "2002-01-01" as all those qualifying statements would be about *the same* triple, so it'd be nonsense.

Instead, according the RDF-star draft, you need to introduce a blank node that identifies *an occurence of* the triple for each role you want to describe.

to complement: 1) You need to introduce a *node* -- it does not have to be blank. 2) The example in the RDF-star spec presents a patter, that can be generalized: different relations between a quoted triple and its occurrences/realizations/roles...