Ideally, this would actively avoid overlaps using some kind of annealing algorithm. Perhaps modeling the distance between the label center and the node center as a spring balanced against repulsion from other labels. Maybe look at how ggplot2 does it.
Ideally, this would actively avoid overlaps using some kind of annealing algorithm. Perhaps modeling the distance between the label center and the node center as a spring balanced against repulsion from other labels. Maybe look at how ggplot2 does it.
This came up in the Google groups forum:
https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/metacoder-discussions/CqtzrnUFOi4/Q_9DVPUgBQAJ