Part of the School of Computer Science’s Information Management Group.
We conduct a series of user studies to extract as much implicit information as possible with respect to users’s visual perception and attention:
Metrics of Visual Complexity is an empirical study to extract more information on how sighted users describe a visually simple or complex Web page. By using card sorting and triadic elicitation techniques we were able to encourage participants to articulate their opinion and extract their implicit knowledge with respect to page presentation. During this study we concluded that the main factors that affect the visual complexity of a Web page are the diversity, density and position of the page elements (e.g. links, words, images). That is, a visually simple page has a significantly less number and variety of page elements than a visually complex page. The most important difference between visually simple and complex pages is the number of different subjects that each present. For example, visually complex Web pages like MSN provide a variety of information to a reader (news, entertainment etc.) where visually simple pages like the Mint Group focus on one subject. Hence, fewer page composition elements are needed to organized a visually simple page than a complex.
In addition, we were able to perform a Pilot Eye Tracking Study. We wanted to investigate how sighted users perceive the visual presentation of Web pages, including on which page areas they glance first and for how long. A descriptive analysis of this study’s results revealed some interesting patterns. For example, Web page salient elements, such as big logos, pictures and animations, attracted the subjects’s attention first but the main content of the page attracted them for the longest. Also, menus were not completely read and no specific reading order between right and left menus or columns was determined. In addition, this pilot study gave implicit information on user’s common interaction scanpath. After gazing at dominant graphics the participants scanned through the main content of the page. They did not fully read the text, in the main content area, and they paid more attention on the links and on the first sentence of each block or paragraph. Then, the participants tended to look and read the first three links on the menu, either on the left or right hand side of the screen. A quantitative analysis could not reveal solid conclusions due to experiment design problems we had, such as participants’s page familiarity and imprecise task assignment. However, we could notice a possible relation between visually complex designs identified by our empirical study and participants’ Web behaviour. For example, participants spent more time interacting with a visually complex page than a simple page. In addition, the more visually complex the page was, the more scattered and disordered the participants’s scanpath was.
Next, we run the Visual Complexity Ranking Experiment, an online experiment, where images were asked to be ranked based on user’s visual perception of complexity. Part of the ViCRAM project is to determine how Web page structural elements (such as font, tables, links, and images) and their characteristics (such as colour and size) can be used to determine the visual presentation and complexity level of a Web page. This experiment investigated how users perceive the visual complexity and aesthetic appearance of Web pages. Results show a strong correlation between users’ visual complexity perception of a Web page, structural elements and aesthetic appearance. Data analysis derived models that estimate the visual complexity and aesthetic level of a Web page using the structural elements of the page. To find out more please read the experiment’s technical report.
The visual appearance of a Web page influences the way a user will interact with the page. A pilot study, showed that the number of different sections a page is organized into affects user perception of the visual complexity level of the page. That is, the more chunks (sections) a Web page is arranged into, the more visually complex it becomes. This Image Comparison Experiment, an online pairwise experiment, extends our pilot study. A set of variables are defined and used to modify a Web page into its chunk rendering. Corner, blocks, boxes and top-left-corners are also defined and identified on a set of Web pages’ chunk renderings. Users had to compare this set of Web pages based on their visual complexity. Results show that the visual complexity score is significantly related with the organisation of the page determined by the number of boxes, blocks, corners and top-left-corners. To find out more please read the experiment’s technical report.
Currently we run the Visual Complexity Ranking Experiment II, an online experiment, where images are asked to be ranked based on user’s visual perception of complexity. This experiment is under progress and results will be available shortly after the experiment is finished. We are seeking for volunteers to take part in the study. To find out more or to take part please visit the Visual Complexity Ranking Experiment II pages.