Human-Centered Educational AI and Learning Analytics
I study human-centered educational AI systems that connect learning analytics, small/local language models, and real-world school data to support teachers, assessment, and classroom decision-making.
My research sits at the intersection of Learning Analytics, Artificial Intelligence in Education, Educational Data Mining, and Human–AI Interaction — with a focus on how AI can support teachers and learners without replacing human judgment, and how educational AI can be deployed responsibly in authentic school and teacher education contexts in Japan.
- Small/Local LLMs × Learning Analytics × Real School Contexts in Japan
- Access to teacher education and school-based research settings
- Japanese educational benchmark datasets & tools (open on GitHub)
- Publications and talks auto-synced from researchmap
About
Who I am
Kyosuke TakamiAssociate Professor
Division of Mathematical and Information Sciences, Osaka Kyoiku University, Japan
I am an educational AI and learning analytics researcher based in Japan. My work focuses on designing, evaluating, and understanding human-centered AI systems for education, with a particular emphasis on small/local language models, real-world school data, teacher support, assessment, and classroom decision-making.
- AI in Education
- Learning Analytics
- Educational Data Mining
- Human–AI Interaction
Research Agenda
Three research threads
Small/Local LLMs for Education
How small and locally deployable language models can support educational feedback, assessment, inquiry learning, and teacher decision-making under real-world constraints such as privacy, cost, transparency, and reliability. Includes benchmark datasets for evaluating language models in Japanese educational contexts.
Learning Analytics & Teacher Support
Methods for analyzing learning data from classrooms, digital platforms, computer-based testing (CBT), and teacher education settings — connecting computational models with meaningful educational interpretation. Includes AI-supported assessment, explainable recommenders, and self-explanation analysis.
Human–AI Classroom Systems
How learners, teachers, AI agents, and classroom social networks interact. This includes AI-supported inquiry learning, classroom social network simulation, and the effects of AI-mediated interventions on learning, motivation, social relationships, and bullying prevention.
Research Vision
My long-term goal is to build a scientific foundation for educational AI systems that are not only technically capable, but also pedagogically meaningful, socially responsible, and usable in real educational institutions. Rather than treating AI as a standalone tool, my work views AI as a new participant in human learning systems — I aim to understand how AI changes the relationships among learners, teachers, data, assessment, and educational decision-making.
Projects / Grants
Auto-generated from researchmap research projects
Publications
Loading…
Datasets & Tools
Open resources on GitHub
HypoGen-LA
Dataset for human evaluation of LLM-generated hypotheses in Learning Analytics research.
github.com/KyosukeTakami/HypoGen_LAJP Center Examination for LLMs
Japanese National University Entrance Examination benchmark for evaluating (local) LLMs.
github.com/KyosukeTakami/center-examination-jpGakucho Benchmark
Japanese national school achievement survey (全国学力・学習状況調査) question set for LLM evaluation.
github.com/KyosukeTakami/gakucho-benchmarkTalks / Presentations
Auto-generated from researchmap presentations
- Loading…
Collaborate
What we can build together
I welcome international collaborations with researchers in Artificial Intelligence in Education, Learning Analytics, Educational Data Mining, Human–AI Interaction, and AI for Education. My lab is particularly well positioned to collaborate on studies that require real educational contexts in Japan — including teacher education, school-based implementation, computer-based testing, learning analytics, and the evaluation of small/local language models for education.
Possible Collaboration Topics
Small & Local LLMs for Education
Comparing small/local language models and frontier models for educational feedback, assessment, inquiry learning, and teacher support.
Privacy-Preserving Learning Analytics
Designing educational AI systems that operate under privacy, institutional, and cost constraints in real schools.
Teacher–AI Collaboration
Studying how teachers interact with AI systems for lesson planning, assessment, classroom analysis, and student support.
Classroom Social Network Simulation
AI-based simulation models of classroom relationships, peer interaction, bullying prevention, and collaborative learning.
Cross-Cultural Educational AI
Comparing how AI systems are perceived, used, and evaluated across educational systems, including Japan and international contexts.
Human-Centered Evaluation
Mixed-methods evaluation of educational AI with teachers, students, and practitioners in authentic settings.
What My Lab Can Contribute
- Access to Japanese teacher education and school-based research contexts
- Experience in learning analytics, educational data science, and AI-supported assessment
- Collaboration with educational institutions and practitioners
- Development and evaluation of small/local LLM systems
- Japanese educational benchmark datasets and tools
- Mixed-methods evaluation combining quantitative data, qualitative insights, and practitioner perspectives
Target Venues
Potential outputs include papers for AIED, LAK, EDM, IJAIED, Computers & Education: Artificial Intelligence, IEEE Transactions on Learning Technologies, and related venues.
Visiting & International Research
Building long-term international partnerships
I am interested in building long-term international collaborations with research groups working on AI in Education, Learning Analytics, Human–AI Interaction, and responsible AI for learning.
A central motivation of my international research agenda is to connect advances in AI with authentic educational settings. Japan provides an important context for studying how educational AI can be designed, evaluated, and governed in real schools, teacher education programs, and public education systems — and I can bring this field site to international partners.
Through international collaboration, I aim to develop comparative and cross-cultural studies on small/local language models, teacher–AI collaboration, learning analytics, AI-supported assessment, and classroom decision-making. I am particularly interested in future visiting research opportunities with groups that share an interest in human-centered educational AI, including research communities around AIED, LAK, EDM, and related international networks.
Contact
Email and researcher profiles