基于知识点关系增强的静态认知诊断模型

doi:10.3969/j.issn.1000-5641.2025.05.008

摘要/Abstract

摘要：

认知诊断作为个性化教育的核心任务, 旨在通过学生历史答题记录评估其对知识点的掌握程度. 现有静态认知诊断模型通常依赖人工标注的关键知识点, 忽视题目中潜在关联的知识点及不同题目对知识点的侧重差异. 提出了一种基于知识点关联关系增强的静态认知诊断模型(Q-matrix Enhanced Neural Cognitive Diagnosis, QENCD), 通过构建知识点依赖关系和题目侧重信息优化题目-知识点关联向量, 并引入残差连接融合两者特征. 实验表明, QENCD模型在ASSIST09、ASSIST17和Junyi数据集上的性能表现均显著优于现有模型.

关键词: 认知诊断, 知识点关联, 注意力机制, 残差连接, 静态模型

Abstract:

Cognitive diagnosis, a core task in personalized education, aims to evaluate students’ mastery of knowledge points using historical response records. Existing static cognitive diagnosis models are typically based on manually annotated key knowledge points, ignoring potential correlations between knowledge points within items as well as differences in how items emphasize specific knowledge points. To address these limitations, this study proposes a static cognitive diagnosis model improved by knowledge point relations (Q-matrix Enhanced Neural Cognitive Diagnosis, QENCD) model. The model optimizes the item-knowledge point association vector by constructing knowledge point dependency relationships and item emphasis information, then integrating these features through residual connections. The experimental results showed that QENCD model significantly outperforms existing models on the ASSIST09, ASSIST17, and Junyi datasets significantly outperforming state-of-the-art baselines. This study provides a more precise knowledge modeling method for static cognitive diagnosis.

Key words: cognitive diagnosis, knowledge point correlation, attention mechanism, residual connection, static model

中图分类号:

G40-034

梁恒贵, 朱益辉, 唐晓雯, 朱命冬. 基于知识点关系增强的静态认知诊断模型[J]. 华东师范大学学报（自然科学版）, 2025, 2025(5): 76-86.

Henggui LIANG, Yihui ZHU, Xiaowen TANG, Mingdong ZHU. Static cognitive diagnosis model enhanced by knowledge point relations[J]. J* E* C* N* U* N* S*, 2025, 2025(5): 76-86.

图/表 7

图1

表1

表2

表3

表4

表5

表6

参考文献 27

1	WANG F, GAO W B, LIU Q, et al. A survey of models for cognitive diagnosis: New developments and future directions [EB/OL]. (2024-07-07) [2025-06-26]. https://arxiv.org/abs/2407.05458.
2	JIANG B, LI X Y, YANG S H, et al.. Data-driven personalized learning path planning based on cognitive diagnostic assessments in MOOCs. Applied Sciences, 2022, 12 (8): 3982- 3982.
3	MENG L L, ZHANG W X, CHU Y, et al.. LD–LP generation of personalized learning path based on learning diagnosis. IEEE Transactions on Learning Technologies, 2021, 14 (1): 122- 128.
4	DU H L, LI N, MA F M, et al. Personalization exercise recommendation based on cognitive diagnosis [C]// The 6th International Conference on Computer Science and Application Engineering. ACM, 2022: 1–5.
5	BI H Y, CHEN E H, HE W D, et al.. BETA-CD: A Bayesian meta-learned cognitive diagnosis framework for personalized learning. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37 (4): 5018- 5026.
6	WANG F, HUANG Z Y, LIU Q, et al.. Dynamic cognitive diagnosis: An educational priors-enhanced deep knowledge tracing perspective. IEEE Transactions on Learning Technologies, 2023, 16 (3): 306- 323.
7	HARVEY R, HAMMER A.. Item response theory. The Counseling Psychologist, 1999, 27 (3): 353- 383.
8	RECKASE M D.. 18 multidimensional item response theory. Handbook of Statistics, 2006, 26, 607- 642.
9	DE LA TORRE J.. DINA model and parameter estimation: A didactic. Journal of Educational and Behavioral Statistics, 2009, 34 (1): 115- 130.
10	LIU Q, WU R Z, CHEN E H, et al.. Fuzzy cognitive diagnosis for modelling examinee performance. ACM Transactions on Intelligent Systems and Technology, 2018, 9 (4): 1- 26.
11	CEN H, KOEDINGER K, JUNKER B.. Learning factors analysis—a general method for cognitive model evaluation and improvement. Intelligent Tutoring Systems, 2006, 4053, 164- 175.
12	PAVLIK P I, HAO C, KOEDINGER K R. Performance factors analysis—a new alternative to knowledge tracing [M]// DIMITROVA V, MIZOGUCHI R, BOULAY B, GRAESSER A, et al. Artificial Intelligence in Education: Building Learning Systems that Care. Amsterdam: IOS Press, 2009.
13	WANG F, LIU Q, CHEN E H, et al.. Neural cognitive diagnosis for intelligent education systems. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34 (4): 6153- 6161.
14	WANG F, LIU Q, CHEN E H, et al.. NeuralCD: A general framework for cognitive diagnosis. IEEE Transactions on Knowledge and Data Engineering, 2022, 35 (8): 8312- 8327.
15	TONG S W, LIU Q, YU R L, et al. Item response ranking for cognitive diagnosis [C]// Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence Organization, 2021: 1750–1756.
16	HUANG W, CHEN E H, LIU Q, et al. Hierarchical multi-label text classification: An attention-based recurrent network approach [C]// Proceedings of the 28th ACM International Conference on Information and Knowledge Management. ACM, 2019: 1051–1060.
17	MIKOLOV T, CHEN K, CORRADO G, et al. Efficient estimation of word representations in vector space [EB/OL]. (2013-09-07) [2025-06-26]. https://arxiv.org/abs/1301.3781.
18	VIE J J, KASHIMA H.. Knowledge tracing machines: Factorization machines for knowledge tracing. Proceedings of the AAAI Conference on Artificial Intelligence, 2019, 33 (1): 750- 757.
19	RENDLE S. Factorization machines [C]// 2010 IEEE International Conference on Data Mining. IEEE, 2010: 995–1000.
20	WANG Z, GU Y, LAN A, et al. VarFA: a variational factor analysis framework for efficient Bayesian learning analytics [EB/OL]. (2020-05-26) [2025-06-26]. https://arxiv.org/abs/2005.13107.
21	FENG M Y, HEFFERNAN N, KOEDINGER K. Addressing the assessment challenge with an online system that tutors as it assesses [J]. User Modeling and User-Adapted Interaction, 2009, 19(3): 243-266.
22	CHANG H, HSU H, CHEN K. Modeling exercise relationships in e-learning: A unified approach [C]// Proceedings of the International Conference on Educational Data Mining. 2015: 532-535.
23	范士青, 刘华山.. 常见的认知诊断模型及其比较. 教育测量与评价(理论版), 2015, (7): 4- 9.
24	SALAKHUTDINOV R, MNIH A. Probabilistic matrix factorization [C]// Proceedings of the 20th International Conference on Neural Information Processing Systems. 2007: 1257–1264.
25	TOSCHER A, JAHRER M. Collaborative filtering applied to educational data mining [J]. Journal of Machine Learning Research: Workshop and Conference Proceedings, 2010, 12.
26	ZHANG Y F, CHEN X.. Explainable recommendation: A survey and new perspectives. Foundations and Trends® in Information Retrieval, 2020, 14 (1): 87- 101.
27	KO H, LEE S, PARK Y, et al.. A survey of recommendation systems: Recommendation models, techniques, and application fields. Electronics, 2022, 11 (1): 141- 154.

数据集	学生数量/人	题目数量/道	知识点数量/个	答题记录条数/条	平均每道题目的正确率	平均每道题目包含的知识点数量/个
ASSIST09	4163	17746	123	324574	0.6587	1.19
ASSIST17	1709	3162	102	311569	0.4365	1.00
Junyi	10000	835	835	220799	0.6516	1.00

模型类型	模型名称	ACC	MSE	AUC
基于统计学的方法	IRT	0.6742	0.4642	0.6852
	MIRT	0.7011	0.4612	0.7192
	DINA	0.6502	0.4673	0.6744
	VFA	0.6811	0.4621	0.6741
基于神经网络的方法	NeuralCDM	0.7198	0.4392	0.7491
基于神经网络的方法	KaNCD	0.7280	0.4289	0.7601
本文提出的方法	QENCD	0.7426	0.4195	0.7754

模型类型	模型名称	ACC	MSE	AUC
基于统计学的方法	IRT	0.6596	0.4653	0.7237
	MIRT	0.6817	0.4648	0.7413
	DINA	0.6484	0.4666	0.6964
	VFA	0.6745	0.4651	0.6894
基于神经网络的方法	NeuralCDM	0.6921	0.4513	0.7534
基于神经网络的方法	KaNCD	0.7155	0.4339	0.7810
本文提出的方法	QENCD	0.7327	0.4243	0.8028

模型类型	模型名称	ACC	MSE	AUC
基于统计学的方法	IRT	0.6760	0.4268	0.7750
	MIRT	0.7513	0.4117	0.7989
	DINA	0.7422	0.4172	0.7872
	VFA	0.6811	0.4421	0.7589
基于神经网络的方法	NeuralCDM	0.7443	0.4178	0.7895
基于神经网络的方法	KaNCD	0.7653	0.4044	0.8186
本文提出的方法	QENCD	0.7789	0.3914	0.8320

控制条件	AUC
控制条件	ASSIST09	ASSIST17	Junyi
QENCD-RK	0.7624	0.7867	0.8087
QENCD-RE	0.7681	0.7925	0.8167
QENCD-RKE	0.7557	0.7808	0.8023
QENCD-RG	0.7663	0.7912	0.8153
QENCD	0.7758	0.8028	0.8320