They are relational databases that provide the strengths of the relational model and its set processing.
Why not just model the data as a document, which doesn't necessarily fit into a rigid relational model but roughly follows the semantics of a high-level model?