Edge Bundling without Reducing the Source to Target Traceability

Paper pdf-Version

Example Graph where the Edge Bundling Technique is Applied

This is the idea of PhD student Fabian Beck, at this time working at the University of Trier in the department of software engineering. The traditional idea when applying edge bundling is to reduce visual clutter by bundling graph edges together that all partially have the same way between source and target vertices. Following this approach - proposed by Danny Holten in 2006 - definitely reduces visual clutter but single edges cannot be traced that easily because source and target information is merged together.

Fabian got the idea to only bundle those edges together that start at the same source vertex OR point to the same target vertex. A pre-defined angle decides when to split an edge from the bundle. If the angle between the straight line pointing to the barycenter of all source/target vertices and the direct and straight connection to the target vertex to be split off gets larger than this pre-defined angle, the corresponding link is actually split off.

We presented this idea as a VisWeek 2011 Poster. The corresponding two page paper can be obtained by clicking on the link above. If you need further details please contact Fabian Beck at the University of Trier.