Dynamic Cognitive Structure Modeling and Explicit Forgetting Computation for Knowledge Tracing

doi:10.3778/j.issn.1673-9418.2412088

Abstract

Abstract: Knowledge tracing task aims to estimate students?? knowledge state based on students?? learning records, in which accurate modeling of learning and forgetting behavior is the key to accurately depict students?? knowledge state. Existing approaches use a static knowledge structure to improve the learning modeling, and improve forgetting modeling by concatenating temporal features and interaction information. However, students?? cognitive structures change over time, and implicitly processing temporal features does not fully utilize temporal information. In order to solve the above problems, this paper proposes a framework called dynamic cognitive structure modeling and explicit forgetting computation for knowledge tracing (CSFKT), to better capture the evolution of students?? knowledge states. The method first updates the adjacency matrix of students?? cognitive structure using a GRU (gated recurrent unit) based on students?? question answering interactions to construct a dynamically changing students?? cognitive structure. Then, based on the cognitive structure, the neighbourhood aggregation is used to model the process of interactions between concepts. Next, a forgetting explicit calculation method is proposed, which uses the interval time and the forgetting curve formula to display the calculation of the memory retention probability and the discounted knowledge state. Moreover, this paper uses a GRU to obtain the knowledge state at the current moment and predict the probability of students?? correct answers. A large number of experiments are conducted on three real datasets, and the results prove that CSFKT can not only model the dynamic cognitive structure but also explicitly model students?? forgetting behaviors, with superior performance and good interpretability.

Key words: knowledge tracing, graph neural network (GNN), gated recurrent unit (GRU), cognitive structure, forgetting curve

摘要： 知识追踪任务旨在根据学生历史学习记录来估计学生的知识掌握程度，其中准确建模学习和遗忘行为是精准刻画学生知识状态的关键。现有方法通常基于学生静态知识结构对学习行为进行建模，并通过拼接时间特征和交互信息的隐式方式以改进遗忘建模。然而，学生的认知结构是随时间变化的，且隐式处理时间特征不能充分利用时间信息。为了解决上述问题，提出了一种认知结构动态建模和遗忘显式计算的知识追踪方法（CSFKT），以更好地捕获学生知识状态变化情况。该方法根据学生答题反馈，使用门控循环单元（GRU）对学生认知结构的邻接矩阵进行更新，构建动态变化的学生认知结构图；基于该认知结构图，利用图神经网络的邻域聚合策略建模知识点相互作用过程；提出一种遗忘显式计算方法，利用间隔时间和遗忘曲线公式显示计算记忆保留概率及衰减后的知识状态，再使用GRU获得当前时刻的知识状态，并预测学生正确作答的概率。在三个真实数据集上进行了大量实验，结果表明CSFKT不仅可以建模动态的认知结构也可以显式建模学生遗忘行为，而且具有优越的性能以及良好的可解释性。

关键词: 知识追踪, 图神经网络（GNN）, 门控循环单元（GRU）, 认知结构, 遗忘曲线

ZHANG Wei, LUO Peihua, LI Zhixin, GONG Zhongwei, SONG Lingling. Dynamic Cognitive Structure Modeling and Explicit Forgetting Computation for Knowledge Tracing[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(10): 2683-2696.

张维, 罗佩华, 李志新, 龚中伟, 宋玲玲. 认知结构动态建模和遗忘显式计算的知识追踪模型[J]. 计算机科学与探索, 2025, 19(10): 2683-2696.

References

[1] PARDO A, JOVANOVIC J, DAWSON S, et al. Using learning analytics to scale the provision of personalised feedback[J]. British Journal of Educational Technology, 2019, 50(1): 128-138.
[2] PIROLLI P, KAIRAM S. A knowledge-tracing model of lear-ning from a social tagging system[J]. User Modeling and User-Adapted Interaction, 2013, 23(2): 139-168.
[3] NAKAGAWA H, IWASAWA Y, MATSUO Y. Graph-based knowledge tracing: modeling student proficiency using graph neural network[C]//Proceedings of the 2019 IEEE/WIC/ACM International Conference on Web Intelligence. New York: ACM, 2019: 156-163.
[4] TONG S W, LIU Q, HUANG W, et al. Structure-based know-ledge tracing: an influence propagation view[C]//Proceedings of the 2020 IEEE International Conference on Data Mining. Piscataway: IEEE, 2020: 541-550.
[5] DUAN Z Y, DONG X X, GU H N, et al. Towards more accurate and interpretable model: fusing multiple knowledge relations into deep knowledge tracing[J]. Expert Systems with Applications, 2024, 243: 122573.
[6] NI Q, WEI T J, ZHAO J B, et al. HHSKT: a learner-question interactions based heterogeneous graph neural network model for knowledge tracing[J]. Expert Systems with Applications, 2023, 215: 119334.
[7] WINSTANLEY M A. Stages in theory and experiment. Fuzzy-structuralism and Piagetian stages[J]. Integrative Psycholo-gical and Behavioral Science, 2023, 57(1): 151-173.
[8] EBBINGHAUS H. Memory: a contribution to experimental psychology[J]. Annals of Neurosciences, 2014, 20(4).
[9] WEERAKODY P B, WONG K W, WANG G J, et al. A review of irregular time series data handling with gated recurrent neural networks[J]. Neurocomputing, 2021, 441: 161-178.
[10] NAGATANI K, ZHANG Q, SATO M, et al. Augmenting knowledge tracing by considering forgetting behavior[C]//Proceedings of the 2019 World Wide Web Conference. New York: ACM, 2019: 3101-3107.
[11] SHEN S H, LIU Q, CHEN E H, et al. Learning process-consistent knowledge tracing[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2021: 1452-1460.
[12] ABDELRAHMAN G, WANG Q. Deep graph memory networks for forgetting-robust knowledge tracing[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(8): 7844-7855.
[13] XU B H, HUANG Z Y, LIU J Y, et al. Learning behavior-oriented knowledge tracing[C]//Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2023: 2789-2800.
[14] PAREJA A, DOMENICONI G, CHEN J, et al. EvolveGCN: evolving graph convolutional networks for dynamic graphs[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(4): 5363-5370.
[15] CHE Z P, PURUSHOTHAM S, CHO K, et al. Recurrent neural networks for multivariate time series with missing values[J]. Scientific Reports, 2018, 8: 6085.
[16] CHEN P H, LU Y, ZHENG V W, et al. Prerequisite-driven deep knowledge tracing[C]//Proceedings of the 2018 IEEE International Conference on Data Mining. Piscataway: IEEE, 2018: 39-48.
[17] GAN W B, SUN Y, SUN Y. Knowledge structure enhanced graph representation learning model for attentive knowledge tracing[J]. International Journal of Intelligent Systems, 2022, 37(3): 2012-2045.
[18] LIU Y, YANG Y, CHEN X, et al. Improving knowledge tracing via pre-training question embeddings[EB/OL]. [2024-10-21]. https://arxiv.org/abs/2012.05031.
[19] XU J, HUANG X Y, XIAO T, et al. Improving knowledge tracing via a heterogeneous information network enhanced by student interactions[J]. Expert Systems with Applications, 2023, 232: 120853.
[20] HE Z Y, LI W, YAN Y H. Modeling knowledge proficiency using multi-hierarchical capsule graph neural network[J]. Applied Intelligence, 2022, 52(7): 7230-7247.
[21] YANG H L, HU S Z, GENG J, et al. Heterogeneous graph-based knowledge tracing with spatiotemporal evolution[J]. Expert Systems with Applications, 2024, 238: 122249.
[22] CHENG K, PENG L Z, WANG P Y, et al. DyGKT: dynamic graph learning for knowledge tracing[C]//Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. New York: ACM, 2024: 409-420.
[23] CORBETT A T, ANDERSON J R. Knowledge tracing: modeling the acquisition of procedural knowledge[J]. User Modeling and User-Adapted Interaction, 1994, 4(4): 253-278.
[24] PIECH C, BASSEN J, HUANG J, et al. Deep knowledge tracing[C]//Advances in Neural Information Processing Systems 28, 2015: 505-513.
[25] ZHANG J N, SHI X J, KING I, et al. Dynamic key-value memory networks for knowledge tracing[C]//Proceedings of the 26th International Conference on World Wide Web. New York: ACM, 2017: 765-774.
[26] GHOSH A, HEFFERNAN N, LAN A S. Context-aware attentive knowledge tracing[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2020: 2330-2339.
[27] WEI L T, LI B, LI Y, et al. Time interval aware self-attention approach for knowledge tracing[J]. Computers and Electrical Engineering, 2022, 102: 108179.
[28] ZHAO W Z, XIA J, JIANG X P, et al. A novel framework for deep knowledge tracing via gating-controlled forgetting and learning mechanisms[J]. Information Processing & Mana-gement, 2023, 60(1): 103114.
[29] HE L L, TANG J T, LI X, et al. Multi-type factors representation learning for deep learning-based knowledge tracing[J]. World Wide Web, 2022, 25(3): 1343-1372.
[30] HUANG T, YANG H L, LI Z, et al. A dynamic knowledge diagnosis approach integrating cognitive features[J]. IEEE Access, 2021, 9: 116814-116829.
[31] HUANG C Q, WEI H J, HUANG Q H, et al. Learning consistent representations with temporal and causal enhancement for knowledge tracing[J]. Expert Systems with Applications, 2024, 245: 123128.
[32] AUSUBEL D G. Cognitive structure and the facilitation of meaningful verbal learning[J]. Journal of Teacher Education, 1963, 14(2): 217-222.
[33] LIU S, LI Q, SHEN X X, et al. Automated discovery of symbolic laws governing skill acquisition from naturally occurring data[J]. Nature Computational Science, 2024, 4(5): 334-345.
[34] LONG T, LIU Y, ZHANG W, et al. Automatical graph-based knowledge tracing[C]//Proceedings of the 15th International Conference on Educational Data Mining, 2022.
[35] WO?NIAK P A, GORZELA?CZYK E J, MURAKOWSKI J A. Two components of long-term memory[J]. Acta Neurobiologiae Experimentalis, 1995, 55(4): 301-305.
[36] 许智宏, 张惠斌, 董永峰, 等. 问题特征增强的知识追踪模型[J]. 计算机科学与探索, 2024, 18(9): 2466-2475.
XU Z H, ZHANG H B, DONG Y F, et al. Question feature enhanced knowledge tracing model[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(9): 2466-2475.
[37] WANG C Y, MA W Z, ZHANG M, et al. Temporal cross-effects in knowledge tracing[C]//Proceedings of the 14th ACM International Conference on Web Search and Data Mining. New York: ACM, 2021: 517-525.
[38] CHOI Y, LEE Y, CHO J, et al. Towards an appropriate query, key, and value computation for knowledge tracing[C]//Proceedings of the 7th ACM Conference on Learning@Scale. New York: ACM, 2020: 341-344.
[39] LEE W, CHUN J, LEE Y, et al. Contrastive learning for knowledge tracing[C]//Proceedings of the ACM Web Conference 2022. New York: ACM, 2022: 2330-2338.
[40] 吴水秀, 罗贤增, 熊键, 等. 知识追踪研究综述[J]. 计算机科学与探索, 2023, 17(7): 1506-1525.
WU S X, LUO X Z, XIONG J, et al. Review on research of knowledge tracking[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(7): 1506-1525.
[41] CUI J J, CHEN Z Y, ZHOU A M, et al. Fine-grained interaction modeling with multi-relational transformer for know-ledge tracing[J]. ACM Transactions on Information Systems, 2023, 41(4): 1-26.
[42] WU Z Y, HUANG L, HUANG Q H, et al. SGKT: session graph-based knowledge tracing for student performance prediction[J]. Expert Systems with Applications, 2022, 206: 117681.
[43] SHEN S H, CHEN E H, LIU Q, et al. Monitoring student progress for learning process-consistent knowledge tracing[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(8): 8213-8227.
[44] YIN Y, DAI L, HUANG Z Y, et al. Tracing knowledge instead of patterns: stable knowledge tracing with diagnostic transformer[C]//Proceedings of the ACM Web Conference 2023. New York: ACM, 2023: 855-864.
[45] MA H, YANG Y, QIN C, et al. HD-KT: advancing robust knowledge tracing via anomalous learning interaction detection[C]//Proceedings of the ACM Web Conference 2024.New York: ACM, 2024: 4479-4488.