Tamara Munzner was doing this with hyperbolic trees back in the mid-90s at Stanford. I remember this because I always thought it was the most sensible interface for big graphs.