I am prototyping a Japanese-language tutor for introductory programming, starting with the interaction from question entry to feedback and logging.
Implemented so far
- A Streamlit interface for C programming questions
- Mock Japanese responses
- Comprehension feedback
- SQLite logging with anonymous IDs
Current limits
LLM integration, RAG, authentication, and reranking are not implemented. This is an early prototype; its educational effectiveness has not been demonstrated.
Why retain logs?
Connecting a question, response, and comprehension feedback can help identify explanations that need improvement. Classroom observations, including confusion about loop conditions and format specifiers, inform this work.
Next steps
Develop evaluation methods for response quality, learning outcomes, self-efficacy, and teacher workload, with appropriate ethics review and consent before using learner data.