I'm not sure whether this write-up adds anything on top of the existing resources, but Jim Blinn, a graphics researcher at Microsoft Research has written up some course notes on tensor diagrams:
http://research.microsoft.com/apps/pubs/default.aspx?id=79791
Update: I suspected that these diagrams might also be lurking in Bayesian networks and other kinds of graphical models used in machine learning. Although I'm not 100% sure about that, it looks like Dan Piponi (aka. sigfpe from "A Neighborhood of Infinity") mentions Bayesian networks as a possible connection during a talk at the 2009 International Conference on Functional Programming.
