Establishing Usability Evaluation Method and Developing the Evaluation System based on Multimodal Information

点击翻译按钮获取中文摘要

This research explores the use of the multimodal data in evaluating usability, including subjective, physiological, eye tracking and physical data, with the aim of improving current evaluation techniques. We have three main conjectures: Conjecture A: Physiological measures can be used to objectively measure user experience. Conjecture B: Combining eye movement and physical data can get closer insights into user behavior, thereby helping to evaluate task performance and understand individual difference. Conjecture C: Physiological measures of experience will correspond to subjective ratings of the experience, but they will not be always consistent.We have presented a series of three experiments to improve current HCI evaluation techniques by incorporating physiological, physical and eye tracking data into traditional methods. The study results showed these multimodal data can be powerful for improving evaluation techniques:(1) Physiological measures can be powerful for quantitatively an … More d objectively evaluating user experience, especially when used in concern with subjective ratings and video analysis. We find evidence that there are different physiological responses across different stress levels in a game environment. These results provide support for Conjecture A.(2) We argue that that eye movements are a reflection of cognitive process, while physical actions are the result of such process ; thus, combining eye and hand movements can provide more information about user behavior and interaction process. These results support the Conjecture B.(3) We also investigate the relationships between traditional usability index and new usability data sources (eye tracking and physiological data). Physiological data can be mirrored in the subjective reports of assessing user experience. These results emphasize that both subjective and physiological evaluation techniques should be used together because they may provide different information of user experience. These results provide support for our Conjecture C. Less