總計畫暨子計畫一：以眼動分析方法探討程式設計學習歷程之個別差異和眼動即時回饋的影響

The main project aims to develop an adapted STEM learning environment and a platform for design teaching adapted learning materials. Using low-price eye-trackers, we developed eye-tracking-based feedback learning systems plus cloud systems to provide the real-time learning feedbacks. Meanwhile, we developed eye-tracking data analysis tools to support such a low-price adapted learning system. So far, during the past two years, we have utilized the Unity programming to implement the real-time eye-tracking modules for gaze tracking and fixation-based index calculations for predefined static and dynamic AOIs. It can also record log data during learning processes. On the other hand, post-hoc eye-tracking data analysis tools have been also developed using the Matlab and Python programming, which can help decrease the real-time-feedback system loadings in the future adapted learning system design. The outcomes of this project have provided the core eye-tracking techniques to support six digital learning studies in Taiwan. The sub-project one aims to use the new eye-tracking system and tools developed by the main project to explore the visual attention process during static multimedia learning environments, such as the reading and comprehension of computer programs. Also this project aims to explore the roles of individual factors such as learning self-efficacy, prior knowledge and abilities as well as learning perspectives play in the learning processes, which would further support the development of the adapted learning system. In the first year, we have conducted two experiments to test the eye-tracking system and tools, one was the interactive visual search experiment and the other was the physics inquiry-based simulation experiment. Both tests successfully demonstrated the applications of the system to track and analyze eye-movement data validly and reliably. During the second year, we have added log data collection into the system function in order to enhance learning analytics function. So far we have finished five pilot tests: Scratch program reading and comprehension experiment, physics refraction phenomena simulation experiment, physics rainfall vs. velocity simulation inquiry experiment, mathematics geometry text-graphic reading comprehension experiment, and the Python vs. Scratch program reading comprehension experiment. Currently, based on the Python pilot-test results, we have constructed the model for the eye-tracking-based feedback for learning the Python programming language. After implementing the feedback module, a final experiment is currently under way to test the effects of the adapted system, which has been postponed due to the COVID-19 pandemic in 20ss20. The pilot results will be presented at the meeting.