计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (2): 99-106.doi: 10.11896/jsjkx.250600002

• 基于图机器学习的教育数据挖掘 • 上一篇    下一篇

DCL-FKT:融合双重对比学习与遗忘机制的个性化知识追踪模型

李春英1,2, 汤志康1, 庄郅玮1, 李文博1, 郭炎熙1, 张晓薇3   

  1. 1 广东技术师范大学计算机科学学院 广州 510665
    2 广东技术师范大学广东省知识产权大数据重点实验室 广州 510665
    3 广东技术师范大学电子与信息学院 广州 510665
  • 收稿日期:2025-06-03 修回日期:2025-09-24 发布日期:2026-02-10
  • 通讯作者: 汤志康(zutang@126.com)
  • 作者简介:(gscy.li@qq.com)
  • 基金资助:
    国家自然科学基金(61807009);广东省普通高校重点领域专项(2023ZDZX1009);大语言模型技术与应用创新专项人才培养计划(粤教高函[2024]9号);广东技术师范大学智能教育联合实验室项目(GSZLGC2023004)

DCL-FKT:Personalized Knowledge Tracing via Dual Contrastive Learning and ForgettingMechanism

LI Chunying1,2, TANG Zhikang1, ZHUANG Zhiwei1, LI Wenbo1, GUO Yanxi1, ZHANG Xiaowei3   

  1. 1 School of Computer Science,Guangdong Polytechnic Normal University,Guangzhou 510665,China
    2 Guangdong Provincial Key Laboratory of Intellectual Property & Big Data,Guangdong Polytechnic Normal University,Guangzhou 510665,China
    3 School of Electronics and Information,Guangdong Polytechnic Normal University,Guangzhou 510665,China
  • Received:2025-06-03 Revised:2025-09-24 Online:2026-02-10
  • About author:LI Chunying,born in 1978,Ph.D,professor,is a distinguished member of CCF(No.19159S).Her main research interests include social networks and human-centered computing,educational big data analysis and mining,andrela-ted areas.
    TANG Zhikang,born in 1978,Ph.D,lecture.His main research interests include knowledge tracing recommender systems and related areas.
  • Supported by:
    National Natural Science Foundation of China(61807009),Key Areas Special Project of Guangdong Provincial Universities(2023ZDZX1009),Talent Training Program for Large Language Model Technology and Application Innovation(Yuejiao Han [2024] No. 9) and Joint Laboratory Project of Intelligent Education,Guangdong Polytechnic Normal University(GSZLGC2023004).

PDF (PC)

95

摘要： 在教育数字化背景下,精准追踪学生的知识掌握程度成为教育数字化中提升教学质量的关键方法之一。知识追踪旨在通过对学生的多种行为数据(如试题作答情况、在线学习时长等)进行分析,评估学生对知识点的掌握程度。已有的知识追踪方法尽管在学生个性化学习表现预测中取得了较好的效果,但仍存在两方面的挑战:1)无法解决教育场景中普遍存在的数据稀疏问题;2)忽视学生知识获取的复杂动态过程,不能有效刻画知识的动态变化与遗忘规律。为突破这些瓶颈,提出一种融合双重对比学习与遗忘机制的个性化知识追踪模型DCL-FKT。该模型利用问题掩码和替换解决数据稀疏问题,并在传统对比学习框架的基础上,创新性地引入特征对比学习模块消除冗余解,提升模型表征效率。同时,结合遗忘门机制,动态模拟人类遗忘曲线,精准捕捉学生知识随时间衰减的非线性变化,实现对学习过程的动态建模。在真实数据集上的对比实验结果表明,该模型在预测准确率等核心指标上均有显著提升,能更精准地反映学生的真实知识水平,为学生在线个性化学习提供可靠支持。

关键词: 智慧教育, 知识追踪, 双重对比学习, 遗忘机制, 表现预测

Abstract: In the context of educational digitalization,accurately tracking students’ knowledge mastery has become one of the key approaches to improving teaching quality.Knowledge tracing seeks to analyze various types of student behavior data－such as responses to questions and online study duration－to evaluate their mastery of specific knowledge points.Although existing approaches have demonstrated good performance in predicting personalized learning behaviors,they still face two major challenges:1) the widespread issue of data sparsity in educational settings;2) the neglect of the complex and dynamic nature of students’ knowledge acquisition,resulting in an inability to effectively capture changes in knowledge and forgetting patterns.To address these challenges,this paper proposes a personalized knowledge tracing model,DCL-FKT,which integrates dual contrastive lear-ning with a forgetting mechanism.The model alleviates data sparsity through question masking and substitution,building upon the traditional contrastive learning framework,it introduces a feature-level contrastive learning module to eliminate redundant representations and enhance modeling efficiency.In addition,by incorporating a forgetting gate mechanism,the model dynamically simulates the human forgetting curve,allowing it to accurately capture the nonlinear decay of students’ knowledge over time and enabling dynamic modeling of the learning process.Experiments conducted on real-world datasets demonstrate that the proposed model achieves significant improvements in core metrics,such as prediction accuracy.It provides a more accurate reflection of students’ actual knowledge levels and offers reliable support for personalized online learning.

Key words: Intelligent education, Knowledge tracing, Dual contrastive learning, Forgetting mechanism, Performance prediction

中图分类号: 

  • G434
[1]ABDELRAHMAN G,WANG Q,NUNES B.Knowledge tra-cing:A survey[J].ACM Computing Surveys,2023,55(11):1-37.
[2]CORBETT A T,ANDERSON J R.Knowledge tracing:Mode-ling the acquisition of procedural knowledge[J].User Modeling and User-Adapted Interaction,1994,4(4):253-278.
[3]ZAKARIA S,MEDHAT M G,ABDELSAMEA M M.AUQantO:Actionable Uncertainty Quantification Optimization in deep learning architectures for medical image classification[J].Applied Soft Computing,2023,146:110666.
[4]RINALDI A M,RUSSO C,TOMMASINO C.Automatic image captioning combining natural language processing and deep neural networks[J].Results in Engineering,2023,18:101107.
[5]PIECH C,BASSEN J,HUANG J,et al.Deep knowledge tracing[C]//Proceedings of the 29th Annual Conference on Neural Information Processing Systems(NeurIPS).2015:505-513.
[6]WANG F,LIU Q,CHEN E,et al.Neural cognitive diagnosis for intelligent education systems[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2020:6153-6161.
[7]ZHANG J,SHI X,KING I,et al.Dynamic key-value memorynetworks for knowledge tracing[C]//Proceedings of the 26th International Conference on World Wide Web(WWW).2017:765-774.
[8]SU Y,LIU Q W,LIU Q,et al.Exercise-enhanced sequential modeling for student performance prediction[C]//Proceedings of the 32nd AAAI Conference on Artificial Intelligence.2018.
[9]SUN J W,ZHOU J P,LIU S Y Y,et al.Interpretable cognitive tracing based on multi-layer attention networks[J].Journal of Computer Research and Development,2021,58(12):2630-2644.
[10]ABDELRAHMAN G,WANG Q.Knowledge tracing with se-quential key-value memory networks[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval.New York:ACM,2019:175-184.
[11]ZHANG L,XIONG X L,ZHAO S Y,et al.Incorporating rich features into deep knowledge tracing[C]//Proceedings of the 4th ACM Conference on Learning @ Scale.New York:ACM,2017:169-172.
[12]VASWANI A,SHAZEER N,PARMAR N,et al.Attention isall you need[C]//Proceedings of the 31st International Confe-rence on Neural Information Processing Systems(NeurIPS).2017:6000-6010.
[13]CHOI Y,LEE Y,CHO J,et al.Towards an appropriate query,key,and value computation for knowledge tracing[C]//Procee-dings of the Seventh ACM Conference on Learning@Scale.2020:341-344.
[14]GHOSH A,HEFFERNAN N,LAN A S.Context-aware at-tentive knowledge tracing[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.2020:2330-2339.
[15]PANDEY S,KARYPIS G.A self-attentive model for knowledge tracing[C]//Proceedings of the International Conference on Educational Data Mining(EDM).2019.
[16]PELÁNEK R.Modeling students’ memory for application inadaptive educational systems[C]//Proceedings of the International Conference on Educational Data Mining(EDM 2015).Madrid,Spain:International Educational Data Mining Society,2015:480-483.
[17]NAGATANI K,ZHANG Q,SATO M,et al.Augmentingknowledge tracing by considering forgetting behavior[C]//Proceedings of The World Wide Web Conference(WWW 2019).New York:ACM,2019:3101-3107.
[18]BAI Y,WANG H,HE J.Blin:A multi-task sequence recommendation based on bidirectional KL-divergence and linear attention[J].Mathematics,2024,12(15):2391.
[19] LIU Y,YANG Y,CHEN X,et al.Improving knowledge tracing via pre-training question embeddings[C]//Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence(IJCAI 2020).2020:1577-1583.
[20] LEE S W,CHOI Y D,PARK J Y,et al.Consistency and Monotonicity Regularization for Neural Knowledge Tracing[J].ar-Xiv:2105.00607,2021.
[21]LEE W,CHUN J,LEE Y,et al.Contrastive learning for know-ledge tracing[C]//Proceedings of the ACM Web Conference 2022.2022.
[22]CHENG K,PENG L,WANG P,et al.DyGKT:Dynamic graph learning for knowledge tracing[C]//Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining.2024.
[23]WU Z,XIONG Y,YU S X,et al.Unsupervised feature learning via non-parametric instance discrimination[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR).2018.
[24]CHEN T,KORNBLITH S,NOROUZI M,et al.A simpleframework for contrastive learning of visual representations[C]//Proceedings of the International Conference on Machine Learning(ICML).PMLR,2020:1597-1607.
[25]HE K,FAN H,WU Y,et al.Momentum contrast for unsupervised visual representation learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).2020:9729-9738.
[26]GAO T,YAO X,CHEN D.SimCSE:Simple contrastive learning of sentence embeddings[J].arXiv:2104.08821,2021.
[27]CLARK K,LUONG M T,LE Q V,et al.ELECTRA:Pre-trai-ning text encoders as discriminators rather than generators[J].arXiv:2003.10555,2020.
[28]ZHANG D,GENG Y,GONG W,et al.RecDCL:Dual contrastive learning for recommendation[C]//Proceedings of the ACM Web Conference 2024.2024:3655-3666.
[29]CHENG W,DU H,LI C,et al.Uncertainty-aware knowledge tracin[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2025:27905-27913.
[30]DOIGNON J P,FALMAGNE J C.Knowledge spaces[M].Berlin:Springer Science & Business Media,2012.
[31]CHEN T,KORNBLITH S,NOROUZI M,et al.A simpleframework for contrastive learning of visual representations[C]//Proceedings of the International Conference on Machine Learning(ICML).PMLR,2020:1597-1607.
[32]ZBONTAR J,JING L,MISRA I,et al.Barlow Twins:Self-supervised learning via redundancy reduction[C]//Proceedings of the International Conference on Machine Learning(ICML).PMLR,2021:12310-12320.
[33]BARDES A,PONCE J,LECUN Y.VICReg:Variance-invari-ance-covariance regularization for self-supervised learning[J].arXiv:2105.04906,2021.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发