Graph Drawing Aesthetics—Created by Users, Not Algorithms

Abstract

Prior empirical work on layout aesthetics for graph drawing algorithms has concentrated on the interpretation of existing graph drawings. We report on experiments which focus on the creation and layout of graph drawings: participants were asked to draw graphs based on adjacency lists, and to lay them out "nicely." Two interaction methods were used for creating the drawings: a sketch interface which allows for easy, natural hand movements, and a formal point-and-click interface similar to a typical graph editing system. We find, in common with many other studies, that removing edge crossings is the most significant aesthetic, but also discover that aligning nodes and edges to an underlying grid is important. We observe that the aesthetics favored by participants during creation of a graph drawing are often not evident in the final product and that the participants did not make a clear distinction between the processes of creation and layout. Our results suggest that graph drawing systems should integrate automatic layout with the user's manual editing process, and provide facilities to support grid-based graph creation.

HOLA: Human-like Orthogonal Network Layout

“…Means in boxes indicate best actual mean rank for that contest. [34,30] and is supported by the positive correlations seen in the "# Outer trees" column in Table 1; R2 users create "aesthetic bend points" we observed, in contrast to previous research, while bends overall were correlated with a poor ranking, nevertheless, certain bend points serving an obvious (to us) aesthetic purpose were present in most of the top-ranked human layouts. This is true for all except Graph 4 in the 'Human 1st' column of Fig.…”

“…Most recently Purchase et al [30] asked participants to draw a graph specified by an adjacency matrix using a graph drawing sketch tool. They found that edge crossings were avoided and that grid-like layouts were preferred.…”

“…We believe that studies in which users are free to layout the network in any way they like are particularly useful in being free of any bias imposed by an existing automatic technique. However, while the few previous such studies [34,12,30] are certainly relevant to understanding what humans like in orthogonal network layout, none considered orthogonal edges. We therefore conducted a humancomposed network layout study specifically designed to identify the factors that humans regard as important for good orthogonal layout.…”

“…Like [12,30] our study had two stages. In the first stage participants were asked to manually layout some small graphs.…”

“…It is not until quite recently that studies such as [34,12,30] returned to first-principles in understanding the requirements for network layout by investigating the kind of node-link diagrams that humans construct by hand. Such "human authored layout" studies have identified aesthetic factors not considered by early algorithm designers or in the previous usability experiments.…”