An Empirical Evaluation of Force-Directed Graph Layout
Force-directed graph layout is a widely used algorithm for the automatic layout of graphs. Little experimental work has been done exploring the behaviour of the algorithm under a variety of conditions. This thesis carries out three large-scale metric-based experiments. The first explores how the core algorithm behaves under changes to initial conditions. The second looks at extending the force-directed layout algorithm with additional forces to reduce overlaps. The third develops a novel symmetry metric for graphs and uses that to explore the symmetries of graphs. This thesis also carries out a user study to show that the differences reported by metrics in the graphs are reflected in a difference in user performance when using graphs for a free-form selection task.