Affiliations: International WIC Institute, Beijing University of Technology, Beijing 100124, P.R. China. zhong@maebashi-it.ac.jp; lusf@bjut.edu.cn; limi_666@bjut.edu.cn
Note: [] Also works: Beijing Key Laboratory of Magnetic Resonance Imaging and Brain Informatics, Beijing 100053, P.R. China
Note: [] Also works: Beijing Key Laboratory of Magnetic Resonance Imaging and Brain Informatics, Beijing 100053, P.R. China Address for correspondence: International WIC Institute, Beijing University of Technology, Beijing 100124, P.R. China Also works: Department of Life Science and Informatics, Maebashi Institute of Technology, Maebashi 371-0816, Japan
Note: [] Also works: Beijing Key Laboratory of Magnetic Resonance Imaging and Brain Informatics, Beijing 100053, P.R. China
Note: [] Also works: Beijing Key Laboratory of Magnetic Resonance Imaging and Brain Informatics, Beijing 100053, P.R. China Also works: The School of Computer and Communication Engineering, Liaoning ShiHua University, Liaoning 113001, P.R. China
Abstract: Lack of understanding of users' underlying decision making process results in the bottleneck of EB-HCI (eye movement-based human-computer interaction) systems. Meanwhile, considerable findings on visual features of decision making have been derived from cognitive researches over past few years. A promising method of decision prediction in EB-HCI systems is presented in this article, which is inspired by the looking behavior when a user makes a decision. As two features of visual decision making, gaze bias and pupil dilation are considered into judging intensions. This method combines the history of eye movements to a given interface and the visual traits of users. Hence, it improves the prediction performance in a more natural and objective way. We apply the method to an either-or choice making task on the commercial Web pages to test its effectiveness. Although the result shows a good performance only of gaze bias but not of pupil dilation to predict a decision, it proves that hiring the visual traits of users is an effective approach to improve the performance of automatic triggering in EB-HCI systems.
