Dr. rer. nat. Dipl.-Inform. Michael Burch
Email: michael.burch@visus.uni-stuttgart.de


VISUS - Institut für Visualisierung und Interaktive Systeme - Stuttgart

30131051835372181011371523532753521


Visual Adjacency Lists for Dynamic Graphs

Paper pdf-Version
A dynamic graph visualized as dynamic adjacency list

A dynamic graph visualized as dynamic adjacency list

The visualization of dynamic graphs has been researched a lot in the past but to this end there is no approach which investigates the visual metaphor of adjacency lists. Those lists are traditionally used for internal data structures for modeling a static graph.

In this work we designed a visualization technique making use of such list based representations which we also extended to show the time-varying behavior of graphs. Also a comparative user study was conducted in order to analyze how the novel representation performs compared to existing techniques. We present a visual representation for dynamic, weighted graphs based on the concept of adjacency lists. Two orthogonal axes are used: one for all nodes of the displayed graph, the other for the corresponding links. Colors and labels are employed to identify the nodes. The usage of color allows us to scale the visualization to single pixel level for large graphs. In contrast to other techniques, we employ an asymmetric mapping that results in an aligned and compact representation of links. Our approach is independent of the specific properties of the graph to be visualized, but certain graphs and tasks benefit from the asymmetry. As we show in our results, the strength of our technique is the visualization of dynamic graphs. In particular, sparse graphs benefit from the compact representation. Furthermore, our approach uses visual encoding by size to represent weights and therefore allows easy quantification and comparison. We evaluate our approach in a quantitative user study that confirms the suitability for dynamic and weighted graphs. Finally, we demonstrate our approach for two examples of dynamic graphs.