OA-Graphs: Orientation Agnostic Graphs for improving the legibility of simple visualizations on horizontal displays

Abstract

Horizontal displays, such as tabletop systems, are emerging as the de facto platform for engaging participants in collaborative tasks. Despite significant efforts in improving the interactivity of information on such systems, very little research has been invested in understanding how groups of people view data visualizations in such environments. Numerous studies introduced different techniques to support viewing visualization for groups of people, such as duplicating or reorienting the visual displays. However, when visualizations compete for pixels on the display, prior solutions do not work effectively.

In this thesis, I explore whether orientation on horizontal displays impacts the legibility of simple visualizations such as graphs. I have found that users are best at reading a graph when it is right side up, and takes them 20% less time than when it is read upside down. The main objective of this thesis was to investigate whether the readability and understandability of simple graphs can be improved. I have introduced the Orientation Agnostic Graph (OA-Graph) which is legible regardless of orientation. The OA-Graph uses a radial layout which has several interesting properties such as implicit orientation, points equidistant to center, and flexible rearrangement. OA-Graphs perform better than graphs that are presented upside down. I have converted several popular types of graphs into their OA counterpart for improved legibility on tabletop systems. Guidelines are presented that describe how other visualizations can be converted to being orientation agnostic.