Future student-simulation work should extend beyond turn-level math dialogue and address history-selection and faithfulness gaps¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The authors lay out forward-looking directions: studying how to include a student's entire learning history in context, potentially "by leveraging interaction embeddings optimized for student simulation or using intelligent context retrieval"; extending simulation beyond dialogue turns to predicting selected options or help requests; examining faithfulness in fully simulated dialogues with new metrics; and testing the task in other domains including programming and language learning. They also plan evaluation on the ASSISTments tutoring chat log dataset to test generalizability beyond their private math data.

Design Implications¶

Context¶

Requirements¶

• New metrics to compare fully simulated tutor-student dialogues against ground-truth dialogues

Constraints¶

• Current evaluation covers a single private dataset in the math domain and only turn-level simulation

Target Learners¶

• Math students, with planned extension to programming and language learners

Target Learning Goals¶

• Faithful simulation of student behavior for evaluating and training LLM tutors

Affordances¶

• History Conditioned Student Simulation Framework

Related Strategies¶

• Future Directions for Dialogue Knowledge Tracing

Examples¶

-

Key Sources¶

• Zhangqi Duan, Shuyan Huang, Alexander Scarlatos, Jaewook Lee, Simon Woodhead, Andrew Lan. (2026). Who Am I? History-Aware Profiles for Student Simulation in Tutoring Dialogues. arXiv preprint. https://arxiv.org/abs/2605.30051