a novel technique for visualizing & analyzing complex graph structures, such as of social networks. the method is designed specifically for graphs that are “multivariate,” i.e., where each node is associated with several attributes. unlike visualizations which
emphasize global graph topology, PivotGraph uses a simple grid-based approach to focus on the relationship between
node attributes & connections.
the graph is rolled up by two different data attributes (x & y-axis). the area of each circle is proportional to the size variable of the node. because of the grid structure of the graph layout, straight edges would almost always have serious occlusion problems. therefore, quadratic curved edges are drawn between nodes, with widths corresponding to their weights.
[(pdf)]