Time spring layout for visualization of dynamic social networks

Common layout techniques for dynamic networks typically either keep node positions static as the graph changes, or operate by “tweening” optimized layouts between adjacent time slices. These techniques can be problematic because (in the first case) there is significant visual “noise” caused by unnecessary edge crossings, and (in the second case) the nodes change so much from one time slice to another that animation is required to display node movement. This paper describes techniques to balance the benefits of keeping node positions relatively static while allowing enough layout adjustment between slices to demonstrate the changing graph. Comparisons are provided against the common layout procedures for a graph with 20 time slices.