AISE
Introduction: Develop AI-based tools to assist in detecting students’ learning conditions and to provide valuable references for researchers and teachers in their instructional practices.
Method: Students’ learning process data (across various domains, such as arguments generated from learning socioscientific issues, group discussion transcripts, problem-solving processes including eye-tracking and handwriting data, or uncertainty in collaborative learning dialogues) are analyzed using supervised learning approaches to train AI systems that can recognize students’ skills.
1. Using AI to Analyze Uncertainty in Collaborative Learning Dialogues
Develop three algorithms for learning process data collection, data analysis, and process data mining. These include a learning process data collection program (web crawler), a dialogue translation tool, and an uncertainty detection algorithm.
Value and Impact:
(1) The international project promotes academic exchange and enhances the global visibility of Taiwan’s scholarship.
(2) Learning process data collection program (web crawler): This program was developed to automatically retrieve, download, and analyze data from the learning platform (Uptale, a VR immersive learning development platform), collecting both individual and group collaborative learning process data. It greatly reduces the time required for manual data organization and minimizes accuracy issues caused by fatigue.
(3) The participating students come from Taiwan, the Netherlands, and Germany. Since the dataset involves multilingual student interactions, AI is employed to automatically translate all dialogues into English for AI modeling.
(4) In traditional classrooms, teachers often struggle to identify students’ uncertainties in group discussions accurately and in real time. By leveraging AI for uncertainty detection, teachers can better monitor students’ learning states, and more importantly, use students’ uncertainties as opportunities to guide their learning.
2. Using Artificial Intelligence to Automatically Detect Students’ Inquiry Skills in Socioscientific Issues
Reference: Zhang, W. X., Lin, J. J. H., & Hsu, Y.-S. (2025). AI-Assisted Assessment of Inquiry Skills in Socioscientific Issue Contexts. Journal of Computer Assisted Learning, 41(1), e13102.
https://doi.org/10.1111/jcal.13102
(SSCI, Q1, IF = 5.1, EDUCATION & EDUCATIONAL RESEARCH, 13/760)
Value and Impact: Traditional assessments of inquiry learning rely heavily on manual grading and teacher observation, which are time-consuming, labor-intensive, and often subjective. By employing AI to analyze students’ argumentation processes when addressing socioscientific issues, it becomes possible to automatically detect their inquiry skills. Applied in classroom settings, this approach can significantly improve the efficiency and objectivity of assessments, thereby enhancing instructional effectiveness.
3. Using Artificial Intelligence with Multimodal Data (Eye Movement and Handwriting) to Evaluate Students’ Problem-Solving Performance
Reference: Lin, J. J. H.* (2024). AI-assisted evaluation of problem-solving performance using eye movement and handwriting. Journal of Research on Technology in Education, 1–25.
doi:10.1080/15391523.2024.2339474
(SSCI, Q1, IF = 5.1, EDUCATION & EDUCATIONAL RESEARCH, 22/760) (*Corresponding Author)
Value and Impact: In traditional mathematics problem-solving instruction, teachers mainly rely on grading to evaluate student performance. However, when the number of students is large, it is challenging to efficiently incorporate process data (e.g., handwriting and eye movement) into diagnostic assessments of learning difficulties and to identify students’ conceptual challenges during problem solving. This study applies AI to train models with multimodal data (eye movement and handwriting) to recognize students’ problem-solving performance. In classroom applications, this approach can reduce teachers’ workload in diagnosing learning difficulties and enable them to dedicate more time to instructional support.
4. Using Artificial Intelligence to Analyze Self-Regulated Learning Strategies
Reference: Wang, C. Y., & Lin, J. J. H.* (2023). Utilizing artificial intelligence to support analyzing self-regulated learning: A preliminary mixed-methods evaluation from a human-centered perspective. Computers in Human Behavior, 144, 107721.
doi:10.1016/j.chb.2023.107721
(SSCI, Q1, IF = 8.957, PSYCHOLOGY, EXPERIMENTAL, 3/91) (*Corresponding Author)