Raphael Menges

One of the main tasks of usability experts is to gain insights into user interaction behavior on the graphical user interface and suggest improvements to the interface. For this purpose, it is necessary to analyze how interface elements are aligned with responses of users, such as mouse or eye movements, clicks or touch interactions. Nowadays, however, interfaces exhibit rich behaviors of their own, choreographing dynamic elements and orchestrating dynamic reactions to user responses rendering common methods for the analysis of interface usability increasingly difficult, if not impossible. To address this problem, we propose a semi-automated approach for determining visual stimuli and delineating them from each other, such that user responses may be properly grouped according to the stimuli the users encountered. For this, we apply the methodology of shot detection on video recordings to user interactions on dynamic interfaces. We execute our visual stimuli discovery on recordings of real-world Web sites and systematically evaluate its outcomes. In particular, the method can reduce the amount of pixel data to be inspected by a usability expert to 3.25 percent of the original videos on average, while representing the content of the dynamic interfaces with a precision of 92 percent. A user-centered evaluation including a online survey with experts substantiate the applicability of the method in discovering visual stimuli accurately while reducing the effort for usability experts.

Video of the talk

26.11.20 - 10:15
via Big Blue Button