We present a general framework for personalized math exercise recommendation that integrates semantic knowledge tracing and model-based reinforcement learning.

Students' drawn scientific sketch evaluation with cognitive theory based multi-agent framework