SA-MGKT: 基于自注意力融合的多图知识追踪方法

doi:10.3969/j.issn.1000-5641.2024.05.003

摘要/Abstract

摘要：

提出了一种基于自注意力融合的多图知识追踪方法(multi-graph knowledge tracing method based on self-attention, SA-MGKT), 旨在通过学生的历史答题数据, 对其知识的掌握程度进行模型化, 并评估其未来学习的表现. 首先, 该方法构建了学生–习题异质图来表示学生–习题的高阶关系, 通过图对比学习技术捕获学生的答题偏好, 并采用3层LightGCN进行图表征学习. 然后, 引入概念关联超图和有向转换图的信息, 通过超图卷积网络和有向图卷积网络进行节点嵌入. 最后, 通过引入自注意力机制, 成功融合了习题序列的内部信息以及多图表征学习所蕴含的潜在信息, 从而显著提升了知识追踪模型的准确性. 实验数据在3个标准数据集上均展现出令人鼓舞的结果, 模型的分类性能得到了大幅提升, 具体表现为相对于基线模型, 在评估指标上分别提高了3.51%、17.91%和1.47%. 这些结果充分验证了融合多图信息和自注意力机制对于增强知识追踪模型性能的有效性.

关键词: 知识追踪, 图对比学习, 自注意力机制

Abstract:

This study proposes a multi-graph knowledge tracing method integrated with a self-attention mechanism (SA-MGKT), The aim is to model students’ knowledge mastery based on their historical performance on problem-solving exercises and evaluate their future learning performance. Firstly, a heterogeneous graph of student-exercise is constructed to represent the high-order relationships between these two factors. Graph contrastive learning techniques are employed to capture students’ answer preferences, and a three-layer LightGCN is utilized for graph representation learning. Secondly, we introduce information from concept association hypergraphs and directed transition graphs, and obtain node embeddings through hypergraph convolutional networks and directed graph convolutional networks. Finally, by incorporating the self-attention mechanism, we successfully fuse the internal information within the exercise sequence and the latent knowledge embedded in the representations learned from multiple graphs, leading to a substantial enhancement in the accuracy of the knowledge tracing model. Experimental outcomes on three benchmark datasets demonstrate promising results, showcasing remarkable improvements of 3.51%, 17.91%, and 1.47% respectively in the evaluation metrics, compared to the baseline models. These findings robustly validate the effectiveness of integrating multi-graph information and the self-attention mechanism in enhancing the performance of knowledge tracing models.

Key words: knowledge tracing, graph contrastive learning, self-attention

中图分类号:

TP183

王畅, 马丹, 许华容, 陈攀峰, 陈梅, 李晖. SA-MGKT: 基于自注意力融合的多图知识追踪方法[J]. 华东师范大学学报（自然科学版）, 2024, 2024(5): 20-31.

Chang WANG, Dan MA, Huarong XU, Panfeng CHEN, Mei CHEN, Hui LI. SA-MGKT: Multi-graph knowledge tracing method based on self-attention[J]. Journal of East China Normal University(Natural Science), 2024, 2024(5): 20-31.

图/表 5

图1

图2

表1

表2

表3

参考文献 30

1	SHEN S, LIU Q, HUANG Z, et al. A survey of knowledge tracing: Models, variants, and applications [EB/OL]. (2024-04-11) [2024-06-19]. https://arxiv.org/pdf/2105.15106.
2	DWIVEDI P, KANT V, BHARADWAJ K K.. Learning path recommendation based on modified variable length genetic algorithm. Education and Information Technologies, 2018, 23 (2): 819- 836.
3	DO P, NGUYEN K, VU T N, et al. Integrating knowledge-based reasoning algorithms and collaborative filtering into e-learning material recommendation system [C]// Future Data and Security Engineering: 4th International Conference. 2017: 419-432.
4	WANG S, WU H, KIM J H, et al. Adaptive learning material recommendation in online language education [C]// Artificial Intelligence in Education. 2019: 298-302.
5	ZHAO C, ZHAO H, HE M, et al. Cross-domain recommendation via user interest alignment [C]// Proceedings of the ACM Web Conference 2023. 2023: 887-896.
6	XU B, HUANG Z, LIU J, et al. Learning behavior-oriented knowledge tracing [C]// Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2023: 2789-2800.
7	曾凡智, 许露倩, 周燕, 等.. 面向智慧教育的知识追踪模型研究综述. 计算机科学与探索, 2022, 16 (8): 1742- 1763.
8	ABDELRAHMAN G, WANG Q, NUNES B.. Knowledge tracing: A survey. ACM Computing Surveys, 2023, 55 (11): 224.
9	KIPF T N, WELLING M. Semi-supervised classification with graph convolutional networks [EB/OL]. (2016-09-29) [2024-06-18]. https://doi.org/10.48550/arXiv.1609.02907.
10	NAKAGAWA H, IWASAWA Y, MATSUO Y. Graph-based knowledge tracing: Modeling student proficiency using graph neural network [C]// IEEE/WIC/ACM International Conference on Web Intelligence. 2019: 156-163.
11	YANG Y, SHEN J, QU Y, et al. GIKT: A graph-based interaction model for knowledge tracing [C]// Machine Learning and Knowledge Discovery in Databases. 2021: 299-315.
12	LUO R, LIU F, LIANG W, et al. DAGKT: Difficulty and attempts boosted graph-based knowledge tracing [C]// International Conference on Neural Information Processing. 2022: 255-266.
13	ZHANG H, BU C, LIU F, et al. APGKT: Exploiting associative path on skills graph for knowledge tracing [C]// Pacific Rim International Conference on Artificial Intelligence. 2022: 353-365.
14	CUI C, YAO Y, ZHANG C, et al.. DGEKT: A dual graph ensemble learning method for knowledge tracing. ACM Transactions on Information Systems, 2024, 42 (3): 78.
15	CORBETT A T, ANDERSON J R.. Knowledge tracing: Modeling the acquisition of procedural knowledge. User Modeling and User-adapted Interaction, 1994, (4): 253- 278.
16	KÄSER T, KLINGLER S, SCHWING A G, et al.. Dynamic bayesian networks for student modeling. IEEE Transactions on Learning Technologies, 2017, 10 (4): 450- 462.
17	CEN H, KOEDINGER K, JUNKER B. Learning factors analysis—A general method for cognitive model evaluation and improvement [C]// International Conference on Intelligent Tutoring Systems. 2006: 164-175.
18	PAVLIK P I, CEN H, KOEDINGER K R. Performance factors analysis—A new alternative to knowledge tracing [C]// Artificial Intelligence in Education. 2009: 531-538.
19	VIE J J, KASHIMA H. Knowledge tracing machines: Factorization machines for knowledge tracing [C]// Proceedings of the AAAI Conference on Artificial Intelligence. 2019: 750-757.
20	PIECH C, BASSEN J, HUANG J, et al. Deep knowledge tracing [C]// Proceedings of the 28th International Conference on Neural Information Processing Systems. 2015: 505-513.
21	KHAJAH M, LINDSEY R V, MOZER M C. How deep is knowledge tracing? [EB/OL]. (2016-04-08) [2024-06-18]. https://doi.org/10.48550/arXiv.1604.02416.
22	LIU Y, YANG Y, CHEN X, et al. Improving knowledge tracing via pre-training question embeddings [EB/OL]. (2020-12-09) [2024-06-18]. https://doi.org/10.48550/arXiv.2012.05031.
23	ZHANG J, SHI X, KING I, et al. Dynamic key-value memory networks for knowledge tracing [C]// Proceedings of the 26th International Conference on World Wide Web. 2017: 765-774.
24	PANDEY S, KARYPIS G. A self-attentive model for knowledge tracing [EB/OL]. (2019-07-16) [2024-06-18]. https://doi.org/10.48550/arXiv.1907.06837.
25	GHOSH A, HEFFERNAN N, LAN A S. Context-aware attentive knowledge tracing [C]// Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2020: 2330-2339.
26	SONG X, LI J, CAI T, et al.. A survey on deep learning based knowledge tracing. Knowledge-Based Systems, 2022, 258, 110036.
27	YE Y, SHAN Z. HGKT: Hypergraph-based knowledge tracing for learner performance prediction [C]// 2023 International Joint Conference on Neural Networks. 2023: 1-9.
28	TONG S, LIU Q, HUANG W, et al. Structure-based knowledge tracing: An influence propagation view [C]// 2020 IEEE International Conference on Data Mining. 2020: 541-550.
29	SONG X, LI J, TANG Y, et al.. JKT: A joint graph convolutional network based deep knowledge tracing. Information Sciences, 2021, 580, 510- 523.
30	HE X, DENG K, WANG X, et al. LightGCN: Simplifying and powering graph convolution network for recommendation [C]// Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. 2020: 639-648.

数据集	学生数量/个	习题数量/个	概念数量/个	概念平均习题数/个	习题平均概念数/个	记录数/条	正确记录占比/%	错误记录占比/%
ASSIST09	8 139	16 891	101	167.24	1.00	274 590	66.16	33.84
ASSIST17	19 697	3 162	102	37.98	1.23	942 816	37.26	62.74
EdNet	7 572	10 795	188	48.73	2.26	225 625	58.99	41.01

类别	模型	AUC/%
类别	模型	ASSIST2009	ASSIST2017	EdNet
序列模型	DKT	73.37	70.70	67.08
序列模型	DKVMN	73.98	66.70	66.75
基于注意力机制的模型	SAKT	74.98	66.48	63.45
基于注意力机制的模型	AKT	75.61	70.34	69.55
基于GNN的模型	GKT	73.96	69.16	67.89
基于GNN的模型	DGEKT	75.84	77.05	69.70
本文所提出的模型	PN-CLKT	74.31	76.48	70.90
本文所提出的模型	SA-MGKT	79.35	94.96	72.37

类别	变体方法	AUC/%
类别	变体方法	ASSIST2009	ASSIST2017	EdNet
单图	Only PNG (PN-CLKT)	74.31	76.48	70.90
	Only DTG	72.39	75.90	69.61
	Only CAHG	74.66	76.10	69.28
双图	DTG + CAHG	75.84	77.05	69.70
	PNG + CAHG	76.40	77.55	69.96
	PNG + DTG	74.65	76.69	69.88
引入自注意力机制	DTG + CAHG & Attention	78.01	92.51	70.42
	PNG + CAHG & Attention	79.12	84.48	72.28
	PNG + DTG & Attention	75.11	91.58	71.16
本文所提出的模型	SA-MGKT	79.35	94.96	72.37

[1]	段志尚, 冉懿, 吕笃良, 祁杰, 钟佳晨, 袁培森. 基于残差网络和深度可分离卷积增强自注意力机制的窃电识别[J]. 华东师范大学学报（自然科学版）, 2023, 2023(5): 193-204.
[2]	班启敏, 吴雯, 胡文心, 林晖, 郑巍, 贺樑. 基于学习者知识和性格的个性化课程推荐[J]. 华东师范大学学报（自然科学版）, 2022, 2022(6): 87-101.
[3]	黄彩蝶, 王昕萍, 陈良育, 刘勇. 基于堆叠门控循环单元残差网络的知识追踪模型研究[J]. 华东师范大学学报（自然科学版）, 2022, 2022(6): 68-78.
[4]	傅裕, 李优, 林煜明, 周娅. 基于自注意力机制的冗长商品名称精简方法[J]. 华东师范大学学报(自然科学版), 2019, 2019(5): 113-122,167.
[5]	刘恒宇, 张天成, 武培文, 于戈. 知识追踪综述[J]. 华东师范大学学报(自然科学版), 2019, 2019(5): 1-15.