A Matrix-Based Visualization for Exploring Dynamic Compound Digraphs

We introduce a matrix-based visualization technique for exploring time-varying directed and weighted graphs. Two overview representations are shown: one for the time-aggregated relations with attached quantitative weighted attributes and one for the results of an automatic dynamic pattern identification algorithm, i.e., relations accompanied by categorical attributes. Apart from a dynamic edge pattern categorization, our tool can also compute graph-specific properties---such as shortest paths or the existence of cliques---and highlight their evolution over time. The visualization method is complemented by interaction techniques that allow the user to navigate, explore, and browse the data, based on the Visual Information Seeking Mantra---overview first, zoom and filter, then details-on-demand. If an additional hierarchical organization of the vertices is available, this is attached to the matrix by vertical and horizontal layered icicle plots allowing one to explore the data on different levels of hierarchical granularity. The usefulness of the tool is demonstrated by applying it to time-varying migration data in the hierarchically structured world.