Perception of Differences in Directed Acyclic Graphs: Influence Factors & Cognitive Strategies

Abstract

Visual comparison of graphs has become widespread in various disciplines. To support this task, it is necessary to generate graph drawings that facilitate the perception of differences. However, while readability of single graphs has been widely investigated, knowledge about the recognition of differences in graphs in general and directed acyclic graphs (DAGs) specifically is quite limited. Towards this end, we conducted an exploratory qualitative study to ascertain which factors influence the perception of changes in DAGs and to better understand the strategies adopted to compare them. Specifically, we investigated pairs of DAGs of varying structure and size (about 50 to 100 nodes) and displayed side-by-side. Based on video recordings and thinking aloud protocols of 22 participants we identified eight factors including edge crossings, shape, symmetry, white space, and density to affect the perception of differences. In addition, we identified three strategies which were used to compare the graphs: layer-by-layer, hierarchical, and a mixture of these two. These strategies were adapted flexibly by the participants based on the graph structure.