In his proof of the Shannon capacity of a graph, Lovasz utilizes a coordinate representation of the pentagon (namely an orthonormal representation). Who first utilized a coordinate representation for finite/infinite graphs for any purpose?

In his proof of the Shannon capacity of a graph, Lovasz utilizes a coordinate representation of the pentagon (namely an orthonormal representation). Who first utilized a coordinate representation for finite/infinite graphs for any purpose? I am thinking a general vector valued representation. And not just a $2$-d representation or a 'fixed' dimensional representation.