知识追踪模型融入遗忘和数据量因素对预测精度的影响
|
摘要
近年来,在线学习的人越来越多,在在线教学过程中教育者需要同时面对更多学习者,不可能了解每一个学习者的知识弱点与问题领域,并据此为学习者提供个性化的学习指导。本研究的目的是及时、准确推断学习者的问题领域,让学习者清楚自身的知识弱点,让教育者更加了解每一个学习者的知识水平,让在线学习系统自动向学习者推荐高效的学习路径和恰当的学习资源。在实验中,分析对比了知识追踪模型及其扩展模型的预测精度,分析了扩展模型使用学习者的所有数据与每个学习者的部分数据的预测准确性差异。结果显示:知识追踪模型可以较好估计学习者知识掌握情况;知识追踪扩展模型的预测精度更好;模型使用学习者部分数据可以获得比使用全部数据更好的预测精度;在学习者学习过程中遗忘时时存在,扩展模型使用部分数据在加快运行进度的同时更有利于精确估计学习者知识水平进而推荐更有效的个性化学习资源与学习路径。
引文
闾汉原,申麟,漆美.2011.基于“态度”的知识追踪模型及集成技术[J].江苏师范大学学报(自然科学版),29(4):54-57.
王卓,张铭.2015.基于贝叶斯知识跟踪模型的慕课学生评价[J].中国科技论文,10(2):241-246.
Chrysafiadi,K.,&Virvou,M.(2013).Student modeling approaches:Aliterature review for the last decade.Expert Systems with Applica?tions,40(11),4715-4729.
Corbett,A.T.,&Anderson,J.R.(1994).Knowledge tracing:modeling the acquisition of procedural knowledge.User Modeling and UserAdapted Interaction,4(4),253-278.
Esichaikul,V.,Lamnoi,S.,&Bechter,C.(2011).Student modeling in adaptive e-learning systems.Knowledge Management&E-Learning:An International Journal,3(3),342-355.
Gong,Y.,Beck,J.E.,&Heffernan,N.T.(2010,June).Comparing knowledge tracing and performance factor analysis by using multiple model fitting procedures.In International conference on intelligent tu?toring systems(pp.35-44).Springer,Berlin,Heidelberg.
Ha,H.,Hong,Y.,Hwang,U.,&Yoon,S.(2018).Intelligent Knowledge Tracing:More Like a Real Learning Process of a Student.arXiv pre?print arXiv:1805.10768.
Hawkins,W.J.,Heffernan,N.T.,&Baker,R.S.(2014,June).Learning Bayesian knowledge tracing parameters with a knowledge heuristic and empirical probabilities.In International Conference on Intelligent Tutoring Systems(pp.150-155).Springer,Cham.
Martori,F.,Cuadros,J.,&González-Sabaté,L.(2015).Direct Estima?tion of the Minimum RSS Value for Training Bayesian Knowledge Tracing Parameters.International Educational Data Mining Society.
Nedungadi,P.,&Remya,M.S.(2015,March).Incorporating forgetting in the personalized,clustered,bayesian knowledge tracing(pc-bkt)model.In 2015 International Conference on cognitive computing and information processing(CCIP)(pp.1-5).IEEE.
Nooraei,B.B.,Pardos,Z.A.,Heffernan,N.T.,&de Baker,R.S.J.(2011).Less is More:Improving the Speed and Prediction Power of Knowledge Tracing by Using Less Data.In EDM(pp.101-110).
Pardos,Z.,&Heffernan,N.(2010,June).Navigating the parameter space of Bayesian Knowledge Tracing models:Visualizations of the convergence of the Expectation Maximization algorithm.In Education?al Data Mining 2010.
Pardos,Z.A.,&Heffernan,N.T.(2010,June).Modeling individualiza?tion in a bayesian networks implementation of knowledge tracing.In International Conference on User Modeling,Adaptation,and Per?sonalization(pp.255-266).Springer,Berlin,Heidelberg.
Sison,R.,&Shimura,M.(1998).Student modeling and machine learn?ing.International Journal of Artificial Intelligence in Education(IJAIED),9,128-158.
Yudelson,M.(2016).Individualizing Bayesian Knowledge Tracing.Are Skill Parameters More Important Than Student Parameters.In EDM(pp.556-561).
(1)MSE可以评价数据的变化程度,其值越小,说明模型拟合实验数据具有更好的预测精确度。
(2)R2取值区间通常在(0,1)之间,0表示还不如不用模型进行预测直接取均值的情况,1表示预测跟真实结果完美匹配的情况。
(3)AUC被定义为ROC曲线下的面积,取值范围在0.5和1之间,只要AUC值大于0.5,就证明该模型具有一定的预测价值。同时,AUC值越接近1,证明模型预测的准确性越好。
(1)使用准确率(ACC)来判断模型预测结果的好坏,但实际样本数据集中经常会出现数据偏斜的情况,即负类样本数大于正类样本数,或正类样本数大于负类样本数,ACC不能脱离测试数据对模型进行评价,所以使用AUC和ACC结合预测。
王卓,张铭.2015.基于贝叶斯知识跟踪模型的慕课学生评价[J].中国科技论文,10(2):241-246.
Chrysafiadi,K.,&Virvou,M.(2013).Student modeling approaches:Aliterature review for the last decade.Expert Systems with Applica?tions,40(11),4715-4729.
Corbett,A.T.,&Anderson,J.R.(1994).Knowledge tracing:modeling the acquisition of procedural knowledge.User Modeling and UserAdapted Interaction,4(4),253-278.
Esichaikul,V.,Lamnoi,S.,&Bechter,C.(2011).Student modeling in adaptive e-learning systems.Knowledge Management&E-Learning:An International Journal,3(3),342-355.
Gong,Y.,Beck,J.E.,&Heffernan,N.T.(2010,June).Comparing knowledge tracing and performance factor analysis by using multiple model fitting procedures.In International conference on intelligent tu?toring systems(pp.35-44).Springer,Berlin,Heidelberg.
Ha,H.,Hong,Y.,Hwang,U.,&Yoon,S.(2018).Intelligent Knowledge Tracing:More Like a Real Learning Process of a Student.arXiv pre?print arXiv:1805.10768.
Hawkins,W.J.,Heffernan,N.T.,&Baker,R.S.(2014,June).Learning Bayesian knowledge tracing parameters with a knowledge heuristic and empirical probabilities.In International Conference on Intelligent Tutoring Systems(pp.150-155).Springer,Cham.
Martori,F.,Cuadros,J.,&González-Sabaté,L.(2015).Direct Estima?tion of the Minimum RSS Value for Training Bayesian Knowledge Tracing Parameters.International Educational Data Mining Society.
Nedungadi,P.,&Remya,M.S.(2015,March).Incorporating forgetting in the personalized,clustered,bayesian knowledge tracing(pc-bkt)model.In 2015 International Conference on cognitive computing and information processing(CCIP)(pp.1-5).IEEE.
Nooraei,B.B.,Pardos,Z.A.,Heffernan,N.T.,&de Baker,R.S.J.(2011).Less is More:Improving the Speed and Prediction Power of Knowledge Tracing by Using Less Data.In EDM(pp.101-110).
Pardos,Z.,&Heffernan,N.(2010,June).Navigating the parameter space of Bayesian Knowledge Tracing models:Visualizations of the convergence of the Expectation Maximization algorithm.In Education?al Data Mining 2010.
Pardos,Z.A.,&Heffernan,N.T.(2010,June).Modeling individualiza?tion in a bayesian networks implementation of knowledge tracing.In International Conference on User Modeling,Adaptation,and Per?sonalization(pp.255-266).Springer,Berlin,Heidelberg.
Sison,R.,&Shimura,M.(1998).Student modeling and machine learn?ing.International Journal of Artificial Intelligence in Education(IJAIED),9,128-158.
Yudelson,M.(2016).Individualizing Bayesian Knowledge Tracing.Are Skill Parameters More Important Than Student Parameters.In EDM(pp.556-561).
(1)MSE可以评价数据的变化程度,其值越小,说明模型拟合实验数据具有更好的预测精确度。
(2)R2取值区间通常在(0,1)之间,0表示还不如不用模型进行预测直接取均值的情况,1表示预测跟真实结果完美匹配的情况。
(3)AUC被定义为ROC曲线下的面积,取值范围在0.5和1之间,只要AUC值大于0.5,就证明该模型具有一定的预测价值。同时,AUC值越接近1,证明模型预测的准确性越好。
(1)使用准确率(ACC)来判断模型预测结果的好坏,但实际样本数据集中经常会出现数据偏斜的情况,即负类样本数大于正类样本数,或正类样本数大于负类样本数,ACC不能脱离测试数据对模型进行评价,所以使用AUC和ACC结合预测。