This direction has been explored. It has its uses but real world applicability is high limited: 1.) it gets tightly bound to one version of its source data 2) it doesn’t deal with the fact that incoming edges to a node can be just as useful and important as their inverses Person3 alumniOf Org2 Vs Org2 alumni Person3 …these are two ways of describing exactly the same state of affairs. 3) this approach generally makes it hard to extend descriptions using external or private vocabulary, unless something like an rdf api is used