DNC: Dynamic Neighborhood Change Faithfulness Metrics

Abstract

Faithfulness metrics measure how faithfully a visualization displays the ground truth information of the data. For example, neighborhood faithfulness metrics measure how faithfully the geometric neighbors of vertices in a graph drawing represent the ground truth neighbors of vertices in the graph. This paper presents a new dynamic neighborhood change (DNC) faithfulness metric for dynamic graphs to measure how proportional the geometric neighborhood change in dynamic graph drawings is to the ground truth neighborhood change in dynamic graphs. We validate the DNC metrics using deformation experiments, demonstrating that it can accurately measure neighborhood change faithfulness in dynamic graph drawings. We then present extensive comparison experiments to evaluate popular graph drawing algorithms using DNC, to recommend which layout obtains the highest neighborhood change faithfulness on a variety of dynamic graphs.