Thesis Details
Impact of Subjective Visual Perception on Automatic Evaluation of Dashboard Design
Using metrics and quantitative design guidelines to analyze design aspects of user interfaces (UI) seems to be a promising way for the automatic evaluation of the visual quality of user interfaces. Since this approach is not able to replace user testing, it can provide additional information about possible design problems in early design phases and save time and expenses in the future. Analyses of used colors or UI layout are the examples of such evaluation. UI designers can use known pixel-based (e.g., Colorfulness) or object-based (e.g., Balance or Symmetry) metrics which measure chosen UI characteristics, based on the raster or structural representation of UI.
The problem of the metric-based approach is that it does not usually consider users' subjective perception (e.g., subjective perception of color and graphical elements located on a screen). Today's user interfaces (e.g., dashboards) are complex. They consist of several color layers, contain overlapping graphical elements, which might increase ambiguity of users' perception. It might be complicated to select graphical elements for the metric-based analysis of UI, so the selection reflects users' perception and principles of a visual grouping of the perceived shapes (as described by Gestalt psychology). Development of objective metrics and design guidelines usually requires a sufficiently large training set of user interface samples annotated by a sufficient number of users.
This thesis focuses on the automatic evaluation of dashboard design. It combines common knowledge about dashboards with the findings in the field of data visualization, visual perception and user interface evaluation, and explores the idea of the automatic evaluation of dashboard design using the metric-based approach. It analyzes chosen pixel-based and object-based metrics. It gathers the experience of users manually segmenting dashboard screens and uses the knowledge in order to analyze the ability of the object-based metrics to distinguish well-designed dashboards objectively. It establishes a framework for the design and improvement of metrics and proposes an improvement of selected metrics. It designs a new method for segmentation of dashboards into regions which are used as inputs for object-based metrics. Finally, it compares selected metrics with user reviews and asks questions suggesting future research tasks.
dashboard, user interface, UX, usability evaluation, metrics, aesthetics, balance, visual perception, subjective perception, Gestalt laws, segmentation
@phdthesis{FITPT906,
author = "Ji\v{r}\'{i} Hynek",
type = "Ph.D. thesis",
title = "Impact of Subjective Visual Perception on Automatic Evaluation of Dashboard Design",
school = "Brno University of Technology, Faculty of Information Technology",
year = 2019,
location = "Brno, CZ",
language = "english",
url = "https://www.fit.vut.cz/study/phd-thesis/906/"
}