Fig. 5
From: A generalized Robinson-Foulds distance for labeled trees
Empirical comparison of the distance inferred for an increasing number of random edit operations (contraction, extensions, and flips), using the classical Robinson-Foulds distance (left) and Contract-Tree algorithm (right). Because the former ignores node labels, it grossly underestimates the actual number of edits. Our algorithm tracks more closely the actual number of edits