Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.16398 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914485097201664 |
|---|---|
| author | Zhang, Ying Cheng, Ningxi Gao, Yizhu Li, Hongmei Shi, Lehong Young, Nicholas Yuan, Geng Zhai, Xiaoming |
| author_facet | Zhang, Ying Cheng, Ningxi Gao, Yizhu Li, Hongmei Shi, Lehong Young, Nicholas Yuan, Geng Zhai, Xiaoming |
| contents | Q-matrices are a cornerstone of theory-driven assessment and learning analytics, making item demands and students' underlying knowledge components and misconceptions explicit and actionable. However, Q-matrices are typically crafted by experts, making them time-consuming to build, prone to subjectivity, and difficult to validate empirically. We propose a framework for human-AI Q-matrix refinement in which large language models (LLMs) generate candidate Q-matrices using structured, misconception-aware prompting, and NeuralCDM provides an empirical evaluation layer to compare candidates based on how well they explain student response data. We apply the framework to a thermodynamics assessment dataset and benchmark locally deployed LLMs against cloud-served models. Results show that iteratively refined LLM-generated Q-matrices can exceed expert-baseline model fit (AUC 0.780 vs. 0.717), and that locally deployed models achieve comparable performance to cloud APIs, supporting privacy-preserving deployment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_16398 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | A Framework for Human-AI Q-Matrix Refinement: A NeuralCDM Evaluation Zhang, Ying Cheng, Ningxi Gao, Yizhu Li, Hongmei Shi, Lehong Young, Nicholas Yuan, Geng Zhai, Xiaoming Computers and Society Artificial Intelligence Q-matrices are a cornerstone of theory-driven assessment and learning analytics, making item demands and students' underlying knowledge components and misconceptions explicit and actionable. However, Q-matrices are typically crafted by experts, making them time-consuming to build, prone to subjectivity, and difficult to validate empirically. We propose a framework for human-AI Q-matrix refinement in which large language models (LLMs) generate candidate Q-matrices using structured, misconception-aware prompting, and NeuralCDM provides an empirical evaluation layer to compare candidates based on how well they explain student response data. We apply the framework to a thermodynamics assessment dataset and benchmark locally deployed LLMs against cloud-served models. Results show that iteratively refined LLM-generated Q-matrices can exceed expert-baseline model fit (AUC 0.780 vs. 0.717), and that locally deployed models achieve comparable performance to cloud APIs, supporting privacy-preserving deployment. |
| title | A Framework for Human-AI Q-Matrix Refinement: A NeuralCDM Evaluation |
| topic | Computers and Society Artificial Intelligence |
| url | https://arxiv.org/abs/2604.16398 |