H2CD: Semantic-Enhanced Heterogeneous Hypergraph Network With Large Language Model for Cognitive Diagnosis

Youheng Bai , Zitao Liu , Mingliang Hou , Teng Guo , Weiqi Luo

CAAI Transactions on Intelligence Technology ›› 2026, Vol. 11 ›› Issue (5) : 1463 -1479.

PDF (2661KB)
CAAI Transactions on Intelligence Technology ›› 2026, Vol. 11 ›› Issue (5) :1463 -1479. DOI: 10.1049/cit2.70158
ORIGINAL RESEARCH
research-article
H2CD: Semantic-Enhanced Heterogeneous Hypergraph Network With Large Language Model for Cognitive Diagnosis
Author information +
History +
PDF (2661KB)

Abstract

Cognitive diagnosis aims to infer learners' knowledge states from their exercise responses, enabling personalised education at scale. Existing methods represent exercises solely by coarse-grained knowledge component annotations, overlooking semantic content and step-level cognitive processes. This causes diagnostic confounding, in which exercises requiring fundamentally different reasoning pathways become indistinguishable in learned representations. Such confounding degrades diagnostic accuracy by conflating distinct learning deficiencies. To address this, we propose H2CD, a semantic-enhanced heterogeneous hypergraph framework that unifies LLM-based cognitive decomposition with structural learning. First, we leverage a large language model to decompose exercises into step-by-step solution paths and extract fine-grained step-level KCs, capturing procedural knowledge beyond predefined annotations. Second, we construct a heterogeneous hypergraph with two types of hyperedges: knowledge hyperedges that connect exercises to their associated multi-granular KCs, preserving compositional knowledge structures, and interaction hyperedges that capture personalised student-exercise relationships. Third, a two-stage hypergraph neural network performs knowledge structure propagation on static exercise-KC relationships and interaction propagation on dynamic student-exercise patterns. Finally, a KC-aware prediction layer projects representations to a learnable diagnostic axis, enabling interpretable mastery assessment at individual KC granularity. Extensive experiments on three real-world educational datasets demonstrate that H2CD outperforms state-of-the-art methods in both AUC and accuracy.

Keywords

AI for education / cognitive diagnosis / hypergraph learning / large language model / smart education

Cite this article

Download citation ▾
Youheng Bai, Zitao Liu, Mingliang Hou, Teng Guo, Weiqi Luo. H2CD: Semantic-Enhanced Heterogeneous Hypergraph Network With Large Language Model for Cognitive Diagnosis. CAAI Transactions on Intelligence Technology, 2026, 11 (5) : 1463-1479 DOI:10.1049/cit2.70158

登录浏览全文

4963

注册一个新账户 忘记密码

References

[1]

M. Barak and S. Z. Wandering, “ A Web-Based Platform for the Creation of Location-Based Interactive Learning Objects,” Computers & Education 62 (2013): 159-170, https://doi.org/10.1016/j.compedu.2012.10.015.

[2]

P. Anderson, What Is Web 2.0?: Ideas, Technologies and Implications for Education , vol. 1 (JISC Technology and Standards Watch, 2007).

[3]

S. Y. Chen and X. Liu, “ An Integrated Approach for Modeling Learning Patterns of Students in Web-Based Instruction: A Cognitive Style Perspective,” ACM Transactions on Computer-Human Interaction 15, no. 1 (2008): 1-28, https://doi.org/10.1145/1352782.1352783.

[4]

Z. Liu, T. Guo, Q. Liang, et al., “ Deep Learning Based Knowledge Tracing: A Review, a Tool and Empirical Studies,” IEEE Transactions on Knowledge and Data Engineering 37, no. 8 (2025): 4512-4536, https://doi.org/10.1109/tkde.2025.3552759.

[5]

Y. Bai, M. Hou, T. Guo, Z. Liu, and W. Luo, “ A Frequency-Aware Mixture of Heterogeneous Experts Framework for Knowledge Tracing,” in Proceedings of the ACM Web Conference 2026, (2026), 7080-7090.

[6]

S. E. Embretson and S. P. Reise, Item Response Theory for Psychologists (Psychology Press, 2013), https://doi.org/10.4324/9781410605269.

[7]

F. Wang, Q. Liu, E. Chen, et al., “ Neural Cognitive Diagnosis for Intelligent Education Systems,” in Proceedings of the AAAI Conference on Artificial Intelligence, the AAAI Conference on Artificial Intelligence, 34, no. 4 (2020): 6153-6161, https://doi.org/10.1609/aaai.v34i04.6080.

[8]

H. Ma, Z. Huang, W. Tang, and X. Zhang, “ Exercise Recommendation Based on Cognitive Diagnosis and Neutrosophic Set,” in 2022 IEEE 25th International Conference on Computer Supported Cooperative Work in Design (CSCWD), (2022), 1467-1472.

[9]

H. Du, N. Li, F. Ma, and T. Palaoag, “ Personalization Exercise Recommendation Based on Cognitive Diagnosis,” in The 6th International Conference on Computer Science and Application Engineering, (2022), 1-5.

[10]

K. Wancham and K. Tangdhanakanond, “ Development of Learning Path Map of Work and Energy for High Schoolers by Using Cognitive Diagnostic Assessment,” Eurasia Journal of Mathematics, Science and Technology Education 19, no. 11 (2023): em2360, https://doi.org/10.29333/ejmste/13812.

[11]

Y. Bai, Z. Liu, T. Guo, M. Hou, and K. Xiao, “ Prerequisite Relation Learning: A Survey and Outlook,” ACM Computing Surveys 57, no. 11 (2025): 1-28, https://doi.org/10.1145/3733593.

[12]

Y. Bai, Y. Zhang, K. Xiao, Y. Lou, and K. Sun, “ A BERT-Based Approach for Extracting Prerequisite Relations Among Wikipedia Concepts,” Mathematical Problems in Engineering 2021, no. 1 (2021): 3510402-3510408, https://doi.org/10.1155/2021/3510402.

[13]

J. De La Torre, “ The Generalized DINA Model Framework,” Psychometrika 76, no. 2 (2011): 179-199, https://doi.org/10.1007/s11336-011-9207-7.

[14]

W. Ma and J. De La Torre, “ GDINA: An R Package for Cognitive Diagnosis Modeling,” Journal of Statistical Software 93 (2020): 1-26, https://doi.org/10.18637/jss.v093.i14.

[15]

J. De La Torre, “ DINA Model and Parameter Estimation: A Didactic,” Journal of Educational and Behavioral Statistics 34, no. 1 (2009): 115-130, https://doi.org/10.3102/1076998607309474.

[16]

F. Wang, Q. Liu, E. Chen, et al., “ NeuralCD: A General Framework for Cognitive Diagnosis,” IEEE Transactions on Knowledge and Data Engineering 35, no. 8 (2022): 8312-8327, https://doi.org/10.1109/tkde.2022.3201037.

[17]

W. Gao, Q. Liu, Z. Huang, et al., “ RCD: Relation Map Driven Cognitive Diagnosis for Intelligent Education Systems,” in The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, (2021), 501-510.

[18]

S. Yang, M. Chen, Z. Wang, et al., “ DisenGCD: A Meta Multigraph-Assisted Disentangled Graph Learning Framework for Cognitive Diagnosis,” Advances in Neural Information Processing Systems 37 (2024): 91532-91559, https://dl.acm.org/doi/10.5555/3737916.3740821.

[19]

J. Hu, W. Xia, X. Zhang, et al., “ Enhancing Sequential Recommendation via LLM-Based Semantic Embedding Learning,” in Companion the ACM Web Conference, (2024), 103-111.

[20]

M. Hong, Y. Xia, Z. Wang, et al., “ EAGER-LLM: Enhancing Large Language Models as Recommenders Through Exogenous Behavior-Semantic Integration,” in The ACM on Web Conference, (2025), 2754-2762.

[21]

S. Huang, Z. Liu, Q. Liu, J. Chen, and Y. Huang, “ Dual-Attentional Time-Aware Fusion Networks for Knowledge Tracing,” Information Fusion 126 (2026): 103573, https://doi.org/10.1016/j.inffus.2025.103573.

[22]

Y.-W. Lee and Y. Sawaki, “ Cognitive Diagnosis Approaches to Language Assessment: An Overview,” Language Assessment Quarterly 6, no. 3 (2009): 172-189, https://doi.org/10.1080/15434300902985108.

[23]

Y. S. Li Hang and Z. Liu, “ Identifying At-Risk K-12 Students in Multimodal Online Environments: A Machine Learning Approach,” in Proceedings of the 13th International Conference on Educational Data Mining, EDM 2020, Fully Virtual Conference, (2020), 113-117, https://doi.org/10.1007/978-981-97-4442-8_9.

[24]

Z. Liu, G. Xu, T. Liu, et al., “ Dolphin: A Spoken Language Proficiency Assessment System for Elementary Education,” in Proceedings of the Web Conference 2020, (2020), 2641-2647.

[25]

J. Feng, Y. Chen, F. Li, A. Sarkar, and M. Zhang, “ How Powerful Are k-Hop Message Passing Graph Neural Networks,” Advances in Neural Information Processing Systems 35 (2022): 4776-4790, https://dl.acm.org/doi/10.5555/3600270.3600615.

[26]

J. Liu and Z. Hou, “ Establishment of Second-Hand Sailboats Price Prediction Model Based on Random Forest and Exploration of Influencing Factors,” in 2023 IEEE 3rd International Conference on Data Science and Computer Application (ICDSCA), (2023), 1337-1342.

[27]

H. Qian, S. Liu, M. Li, B. Li, Z. Liu, and A. Z. Orcdf, “ An Oversmoothing-Resistant Cognitive Diagnosis Framework for Student Learning in Online Education Systems,” in The 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, (2024), 2455-2466.

[28]

H. Luo, E. Haihong, G. Chen, et al., “ HyperGraphRAG: Retrieval-Augmented Generation via Hypergraph-Structured Knowledge Representation,” in Advances in Neural Information Processing Systems, vol. 38 (2025), 152206-152234.

[29]

J. Shen, H. Qian, S. Liu, W. Zhang, B. Jiang, and A. Zhou, “ Capturing Homogeneous Influence Among Students: Hypergraph Cognitive Diagnosis for Intelligent Education Systems,” in The 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, (2024), 2628-2639.

[30]

S. Huang, H. Li, Y. Gu, X. Hu, Q. Li, and G. Xu. HyperG, “ Hypergraph-Enhanced LLMs for Structured Knowledge,” in Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, (2025), 1218-1228.

[31]

J. Stamper, R. Xiao, and X. Hou, “ Enhancing LLM-Based Feedback: Insights From Intelligent Tutoring Systems and the Learning Sciences,” in International Conference on Artificial Intelligence in Education, (2024), 32-43.

[32]

T. Wang, Y. Zhan, J. Lian, et al., “ LLM-Powered Multi-Agent Framework for Goal-Oriented Learning in Intelligent Tutoring System,” in Companion the ACM on Web Conference, (2025), 510-519.

[33]

Y. Zheng, X. Li, Y. Huang, et al., “ Automatic Lesson Plan Generation via Large Language Models With Self-Critique Prompting,” in International Conference on Artificial Intelligence in Education, (2024), 163-178.

[34]

Y. Liu, S. Bhandari, and Z. A. Pardos, “Leveraging LLM Respondents for Item Evaluation: A Psychometric Analysis,” British Journal of Educational Technology 56, no. 3 (2025): 1028-1052, https://doi.org/10.1111/bjet.13570.

[35]

Q. Jia, J. Cui, H. Du, et al., “ LLM-Generated Feedback in Real Classes and Beyond: Perspectives From Students and Instructors,” in The 17th International Conference on Educational Data Mining, (2024), 862-867.

[36]

C. Hankeln, “ Challenges in Using ChatGPT for Assessing Conceptual Understanding in Mathematics Education,” Journal of Mathematics Education 17, no. 1 (2024): 1-15, https://doi.org/10.26711/007577152790171.

[37]

R. Meissner, A. Pögelt, K. Ihsberner, et al., “ LLM-Generated Competence-Based E-Assessment Items for Higher Education Mathematics: Methodology and Evaluation,” Frontiers in Education 9 (2024): 1427502, https://doi.org/10.3389/feduc.2024.1427502.

[38]

Z. Shao, S. Yuan, L. Gao, Y. He, D. Yang, and S. Chen, “ Unlocking Scientific Concepts: How Effective Are LLM-Generated Analogies for Student Understanding and Classroom Practice?,” in The 2025 CHI Conference on Human Factors in Computing Systems, (2025), 1-19.

[39]

D. Armfield, E. Chen, A. Omonkulov, et al., “ Avalon: A Human-in-the-Loop LLM Grading System With Instructor Calibration and Student Self-Assessment,” in International Conference on Artificial Intelligence in Education, (2025), 111-118.

[40]

O. Fagbohun, N. P. Iduwe, M. Abdullahi, A. Ifaturoti, and O. Nwanna, “ Beyond Traditional Assessment: Exploring the Impact of Large Language Models on Grading Practices,” Journal of Artificial Intelligence, Machine Learning and Data Science 2, no. 1 (2024): 1-8, https://doi.org/10.51219/jaimld/oluwole-fagbohun/19.

[41]

Z. Dong, J. Chen, and F. Wu, “ Knowledge Is Power: Harnessing Large Language Models for Enhanced Cognitive Diagnosis,” AAAI Conference on Artificial Intelligence 39, no. 1 (2025): 164-172, https://doi.org/10.1609/aaai.v39i1.31992.

[42]

Y. Feng, C. Yang, X. Hou, et al., “ Beyond Graphs: Can Large Language Models Comprehend Hypergraphs?,” in The Thirteenth International Conference on Learning Representations, (2025), 1-28.

[43]

Z. Shu, X. Sun, and H. Cheng, “ When LLM Meets Hypergraph: A Sociological Analysis on Personality via Online Social Networks,” in The 33rd ACM International Conference on Information and Knowledge Management, (2024), 2087-2096.

[44]

H. Luo, E. Haihong, Z. Tang, et al., “ Chatkbqa: A Generate-Then-Retrieve Framework for Knowledge Base Question Answering With Fine-Tuned Large Language Models,” in Findings of the Association for Computational Linguistics: ACL 2024, (2024), 2039-2056.

[45]

P. Veličković, G. Cucurull, A. Casanova, A. Romero, P. Lio, and Y. Bengio, “ Graph Attention Networks,” arXiv preprint arXiv:1710.10903 (2017).

[46]

Z. Huang and Z. L. Hcgkt, “ Hierarchical Contrastive Graph Knowledge Tracing With Multi-Level Feature Learning,” in International Conference on Artificial Intelligence in Education (Springer, 2025), 274-288.

[47]

M. D. Reckase, “ Multidimensional Item Response Theory,” Handbook of Statistics 26 (2006): 607-642, https://doi.org/10.1007/978-0-387-89976-3.

[48]

S. Liu, J. Shen, H. Qian, and A. Zhou, “ Inductive Cognitive Diagnosis for Fast Student Learning in Web-Based Intelligent Education Systems,” in The ACM Web Conference, (2024), 4260-4271.

[49]

N. Reimers and I. Gurevych, “ Sentence-Bert: Sentence Embeddings Using Siamese Bert-Networks,” in Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP, 2019), 3982-3992.

PDF (2661KB)

0

Accesses

0

Citation

Detail

Sections
Recommended

/

〈 〉