Just read an interesting paper, “Teaching Model Driven Engineering from a Relational Database Perspective,” the draft of Batory, Latimer, & Azanza. , where stated that “MDE can be understood as an application of relational databases.” And “all graphs [of model or metamodel] have simple encodings as a set of normalized tables.”
This is fully in line with my explanation for MDSD, and the later Model-Driven Applications vs. Model-Driven Engineering. One can say that explanation have no contribution… may be, I just see that somewhat an additional acquisition and as an instance for the universality of MDM.
Their key step is transforming the model and metamodel in to relational database, than implement the MDE style operations (e.g., model transformation) on this basis. Don’t know if that is seen as non-mainstream way, but in my humble eyes, that is indeed meaningful. As mentioned in the paper they are inspired by J. M. Favre (2004) work. Favre said, “The set theory and language theory are parts of the lingua franca in Computer Science, so using these concepts can greatly help in identifying bridges with existing TSs.”  This was a great thought but if did it get the deserved attention? It seems somewhat just languages there, everything in languages… and on the other hand, the really significant and biggest space models and modeling for computer/software application is not belonged MDE/MDSD, i think…
 Batory, D., Latimer, E. & Azanza, M. (2013). Teaching Model Driven Engineering from a Relational Database Perspective. Draft that accepted by MODELS 2013. Retrieved at ftp://ftp.cs.utexas.edu/pub/predator/MDELiteModels13.pdf
 Favre, J.M. (2004). Towards a basic theory to model model driven engineering. In: Workshop on Software Model Engineering, WISME 2004