“…However, since G pag and G usr share the whole vertex set, the centrality of the shared vertices (considered in [16]) is not a meaningful aesthetic requirement in our case. Also, motivated by recent cognitive experiments, we do not insist on minimizing the number of edge crossings (as done in [7,23]) but rather we try to control the visual quality of the edge crossings. More precisely, human-computer interaction experiments have shown that orthogonal crossings do not inhibit human task performance when reading a drawing and that users have a geodesic tendency when discovering relations between pairs of vertices [20,21,22].…”