reasoning: more specialized inference constructs

[VladimirAlexiev]

Continuing #351:

https://graphdb.ontotext.com/documentation/11.0/rules-optimisations.html#propchain gives another important example:
a 2-place PropChain instead of the general owl:propertyChainAxiom.
owl:propertyChainAxiom needs to unroll the rdf:List representing the chain, which makes it harder to implement.
Most chains found in practice are 2-place chains (and a chain of any length can be implemented as a sequence of 2-place chains), so let's consider

PropChain(?p,?q,?r). ?x ?p ?y. ?y ?q ?z => ?x ?r ?z

https://rawgit2.com/VladimirAlexiev/my/master/pubs/extending-owl2/index.html gives yet more variations on the topic, eg
transitiveLeft is like transitiveOver, but extends the chain on the left not the right, i.e.

transitiveLeft(?p,?q). ?x ?q ?y. ?y ?p ?z => ?x ?p ?z
transitiveOver(?p,?q). ?x ?p ?y. ?y ?q ?z => ?x ?p ?z

https://graphdb.ontotext.com/documentation/11.0/rules-optimisations.html#translating-owl-constructs-to-specialized-property-constructs
shows how owl:TransitiveOver and owl:propertyChainAxiom can be converted to these specialized constructs.

Now, what does that mean for SHACL rules? I don't yet know, but it's in the vein of "special constructs can sometimes be implemented more efficiently than more generic or "standard" constructs".