We investigate the problem of visually encoding time-varying weighted digraphs to provide an overview about dynamic graphs. Starting from a rough overview of dynamic relational data an analyst can subsequently explore the data in more detail to gain further insights. To reach this goal we first map the graph vertices in the graph sequence to a common horizontal axis. Edges between vertices are represented as stacked horizontal and color-coded links starting and ending at their corresponding start and end vertex positions. The direction of each edge is indicated by placing it either above or below the horizontal vertex line. We attach a vertically aligned timeline to each link to show the weight evolution for those links. The order of the vertices and stacked edges is important for the readability of the visualization. We support interactive reordering and sorting in the vertex, edge, and timeline representations. The usefulness of our edge-stacked timelines is illustrated in a case study showing dynamic call graph data from software development.
Edge stacking allows compact representations of node-link diagrams by using parallel stacked lines for the graph edges. Color coding visually encodes the weights. The question is if such a visualization which is free of link crossings can improve user task performance compared to traditional approaches.