Abstract
The aim of this research is to establish whether manipulating graphical properties of LineSets affects user performance. LineSets are a technique for visualizing networks of data items which lie in overlapping sets. Information about sets isrepresented by drawing one line for each set over an existing network of data items. Despite a wealth of knowledge of perceptual and cognitive theories being readily available, little work has been done to understand how manipulating graphical properties of LineSets affects user performance. In turn, little guidance is available to visualization designers for effectively visualising data for different types of tasks. Therefore, to draw more effective visualizations, an understanding of the impact of altering graphical properties on task performance is needed.
There are six graphical choices to which humans are known to be perceptually sensitive: size, value, texture, colour, orientation and shape. The effectiveness of these choices is dependent on the type of information that is conveyed and task to be performed. As a result, it seems important to bridge the gap between the
graphical choices that are made and the tasks that users perform. This would allow for more effective visualizations to be drawn if they can be optimised for a type of task. To determine the effect of graphical manipulations in LineSets, a series of empirical experiments that measured participant task completion time
and accuracy were performed.
Our first experiment, which investigated the effect of the set-line colour on user performance, established that applying unique colour hues to the set-lines was the most effective colour treatment regardless of the type of task being performed. The effects of varying the thickness of the set-lines, when the thickness reflects the number of items in a given set, was then evaluated. The results suggested that varying set-line thickness improved task performance. Similar results were found in a further experiment where we varied the diameter of the nodes to reflect the number of incident edges.
A final experiment to evaluate the impact of these changes as a whole, comparing automatically generated LineSets to those thought to be improved by our results, provides evident that adopting our graphical choices significantly improves performance. Therefore, our contributions give insight to both designers and users of set visualisations into which graphical choices to make when conveying information in order to generate diagrams that can be more accurately and quickly interpreted.
Date of Award | May 2019 |
---|---|
Original language | English |
Awarding Institution |
|
Supervisor | Gem Stapleton (Supervisor), James Burton (Supervisor) & Peter Chapman (Supervisor) |