Towards a Vertex and Edge Label Aware Force Directed Layout Algorithm

Many automatic graph layout algorithms can cause shaped vertices and edge labels (which have a size when drawn on the screen) to overlap in the resulting visualisation. Overlaps can hide information that users expect to see in cases where the graph is small. We perform two experiments on a large real-world set of small (10-110 vertex) graphs to compare how different combinations of forces in Eades’ force directed layout algorithm affect the final graph layout. We identify an optimal combination of forces from those we tested. In particular, we found that adding charged walls, variable node charge and edge label charges, minimises overlaps. We also found that using Hooke’s Law over Eades’ logarithmic attractive force tends to reduce edge crossings.