Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
AI Research
arXiv CS.AI·25d ago·5 min read
Enhancing knowledge tracing robustness for new question cold start in Intelligent Tutoring Systems
TL;DR
Practical Integrated Cross-consistent Knowledge Tracing (PICKT) enhances the robustness of knowledge tracing models in Intelligent Tutoring Systems (ITS) when new questions are introduced. Key findings indicate that difficulty, text, and knowledge map features significantly improve diagnostic reliability for unseen questions.
✦ Why It Matters
Prioritize feature annotation in your KT models to improve performance with new questions in educational applications.
Key Takeaways
- Implement PICKT to enhance KT model robustness.
- Focus on difficulty, text, and knowledge map features.
- Analyze question characteristics for better feature annotation.
- Leverage training data to estimate unseen question representations.
How It Works
PICKT integrates multiple feature types to improve knowledge tracing during cold starts. It uses the difficulty of questions to inform the model about challenging items, while text and knowledge map features allow the model to relate new questions to previously seen ones, enhancing its ability to predict learner responses.