Testing and evaluation are essential for refining interface design, but it’s often time-consuming for recruiting human participants and conducting the experiment. Thus, this study aims to simulate the user's visual attention and operation intention during the interaction with an interface, so as to automatically represent the user’s behavior in response to any interface design. It would help not only reduce the difficulties associated with usability testing but also speed up the process of interface development. Focusing on the two-dimensional graphical user interface, prediction models were constructed based on the cognitive processes of human beings. More specifically, the N-SEEV model was adopted for the simulation of user's gaze trajectory, whereas the decision model based on similarity comparison was used for estimating the user's click sequence. Besides, an experiment was designed to validate the simulation results. The ScanMatch algorithm and click perplexity were used to quantify the difference between the "predicted behavior" and the "actual behavior." Both measures were then compared against the prespecified criterion, suggesting that the predicted gaze trajectory and click sequence of the interface are both acceptable. By further integrating the predicted click sequence with predictive models of movement time, user performance could be estimated for the rapid evaluation of interface design.

Authors:
Jun-Ming Lu, National Tsing Hua University, Taiwan
Tsai-Hsuan Lin, National Tsing Hua University, Taiwan

About the Presenter(s)
Dr Jun-Ming Lu is a University Assistant Professor/Lecturer at National Tsing Hua University in Taiwan

See this presentation on the full schedule – Thursday Schedule

Bookmark