An Eye Tracking Study into the Effects of Graph Layout

arXiv:0810.4431v1 [cs.HC] 24 Oct 2008 An Eye Tracking Study into the Effects of Graph Layout Weidong Huang∗ School of Information Technologies, University of Sydney, Australia IMAGEN Program, National ICT Australia Ltd. ABSTRACT Graphs are typically visualized as node-link diagrams. Although there is a fair amount of research focusing on crossing minimiza- tion to improve readability, little attention has been paid on how to handle crossings when they are an essential part of the final visu- alizations. This requires us to understand how people read graphs and how crossings affect reading performance. As an initial step to this end, a preliminary eye tracking experi- ment was conducted. The specific purpose of this experiment was to test the effects of crossing angles and geometric-path tendency on eye movements and performance. Sixteen subjects performed both path search and node locating tasks with six drawings. The re- sults showed that small angles can slow down and trigger extra eye movements, causing delays for path search tasks, whereas crossings have little impact on node locating tasks. Geometric-path tendency indicates that a path between two nodes can become harder to fol- low when many branches of the path go toward the target node. The insights obtained are discussed with a view to further confirmation in future work. Keywords: eye tracking, edge crossing, geometric path, evalua- tion, graph drawing Index Terms: H.1.2 [Models and Principles]: User/Machine Systems—Human Factors; H.5.0 [Information Interfaces and Pre- sentation]: User Interfaces—Evaluation/methodology 1 INTRODUCTION Graphs are typically visualized as node-link diagrams. A graph can be drawn in many different ways by simply changing the layout of nodes. A growing number of empirical studies have shown that graph layout affects not only readability, but also the understanding of the underlying data. In particular, edge crossings (or link cross- ings) has long been a major concern in graph drawing; it is com- monly accepted and employed as a general rule that the number of crossings should be reduced as much as possible [10]. However, in practice, crossing minimization is a hard problem in designing algo- rithms for graph drawing [3]. There are also many graphs in which crossings are not removable. Although there is a fair amount of research focusing on crossing minimization (e.g., [2, 9, 15]) in the literature, little attention has been paid on how to handle crossings when they are an essential part of the final visualizations. Some researchers have pointed out that different crossing styles may have different degrees of impact. Take the two drawings in Figure 1, as an example. These two drawings were of a graph and drawn using two different approaches: k-planarization [15] and minimal-crossing-number [9], respectively. The drawing in Fig- ure 1(a) has 34 crossings, which is 41% more crossings than the drawing in Figure 1(b) has (24 crossings). However, as indicated ∗e-mail: weidong.huang@nicta.com.au (a) (b) Figure 1: Two drawings of the same graph. (a) k-planarization draw- ing, (b) minimal-crossing-number drawing. Adopted from [15, Figure 2] in [15], an informal evaluation revealed that the former drawing was considered as having less crossings and being more readable. Fur- ther, not only the collective crossing pattern has a role in affecting graph perception, but also the individual crossing angle. For exam- ple, as mentioned in [7, 20], when edges cross at nearly-90-degree angles, they are less likely to be confusing than when crossing at acute angles. In addition, more and more empirical studies are available show- ing that in some situations, crossings may not be as bad as we nor- mally think (e.g., [7, 8]). For example, in perceiving sociograms (node-link diagrams for social networks), it was found that cross- ings are important only for tasks that involve path tracing [8]. Even when sociograms are drawn to convey specific information, such as how many social groups there are in the network, it is more desirable to cross edges connecting the group members [8]. It is also possible that drawing graphs without crossings can make some structural features less apparent, such as symmetry. Thus, when the cost of crossing reduction cannot be justified, or when crossings become unavoidable, the questions arise: How can we reduce the negative impact of crossings to the minimum? In what situations can we simply ignore the presence of crossings, or even make use of them? To answer these questions, we need to have knowledge of how and when crossings, or visual layouts in a broader sense, affect graph understanding. In addition, it is also essential for us to have a good understanding of how people read graphs. 1.1 Related Work User studies investigating layout effects can be divided into two groups according to the graphs used: abstract graphs and domain graphs (such as sociograms, UML diagrams). Purchase [17] conducted a user study examin

…(Full text truncated)…

This content is AI-processed based on ArXiv data.