In the field of intelligent education,the integration of artificial intelligence,especially deep learning technologies,has garnered significant attention.Knowledge tracing(KT)plays a pivotal role in this field by pred...In the field of intelligent education,the integration of artificial intelligence,especially deep learning technologies,has garnered significant attention.Knowledge tracing(KT)plays a pivotal role in this field by predicting students’future performance through the analysis of historical interaction data,thereby assisting educators in evaluating knowledgemastery and tailoring instructional strategies.Traditional knowledge tracingmethods,largely based on Recurrent Neural Networks(RNNs)and Transformer models,primarily focus on capturing long-term interaction patterns in sequential data.However,these models may neglect crucial short-term dynamics and other relevant features.This paper introduces a novel approach to knowledge tracing by leveraging a pure Multilayer Perceptron(MLP)architecture.We proposeMixerKT,a knowledge tracing model based on theHyperMixer framework,which uniquely integrates global and localMixer feature extractors.This architecture enables more effective extraction of both long-terminteraction trends and recent learning behaviors,addressing limitations in currentmodels thatmay overlook these key aspects.Empirical evaluations on twowidely-used datasets,ASSIS Tments2009 and Algebra2005,demonstrate that MixerKT consistently outperforms several state-of-the-art models,including DKT,SAKT,and Separated Self-Attentive Neural Knowledge Tracing(SAINT).Specifically,MixerKT achieves higher prediction accuracy,highlighting its effectiveness in capturing the nuances of learners’knowledge states.These results indicate that our model provides a more comprehensive representation of student learning patterns,enhancing the ability to predict future performance with greater precision.展开更多
Knowledge tracing is the key component in online individualized learning,which is capable of assessing the users'mastery of skills and predicting the probability that the users can solve specific problems.Availabl...Knowledge tracing is the key component in online individualized learning,which is capable of assessing the users'mastery of skills and predicting the probability that the users can solve specific problems.Available knowledge tracing models have the problem that the assessments are not directly used in the predictions.To make full use of the assessments during predictions,a novel model,named deep knowledge tracing embedding neural network(DKTENN),is proposed in this work.DKTENN is a synthesis of deep knowledge tracing(DKT)and knowledge graph embedding(KGE).DKT utilizes sophisticated long short-term memory(LSTM)to assess the users and track the mastery of skills according to the users'interaction sequences with skill-level tags,and KGE is applied to predict the probability on the basis of both the embedded problems and DKT's assessments.DKTENN outperforms performance factors analysis and the other knowledge tracing models based on deep learning in the experiments.展开更多
With the popularity of online learning in educational settings, knowledge tracing(KT) plays an increasingly significant role. The task of KT is to help students learn more effectively by predicting their next mastery ...With the popularity of online learning in educational settings, knowledge tracing(KT) plays an increasingly significant role. The task of KT is to help students learn more effectively by predicting their next mastery of knowledge based on their historical exercise sequences. Nowadays, many related works have emerged in this field, such as Bayesian knowledge tracing and deep knowledge tracing methods. Despite the progress that has been made in KT, existing techniques still have the following limitations: 1) Previous studies address KT by only exploring the observational sparsity data distribution, and the counterfactual data distribution has been largely ignored. 2) Current works designed for KT only consider either the entity relationships between questions and concepts, or the relations between two concepts, and none of them investigates the relations among students, questions, and concepts, simultaneously, leading to inaccurate student modeling. To address the above limitations,we propose a graph counterfactual augmentation method for knowledge tracing. Concretely, to consider the multiple relationships among different entities, we first uniform students, questions, and concepts in graphs, and then leverage a heterogeneous graph convolutional network to conduct representation learning.To model the counterfactual world, we conduct counterfactual transformations on students’ learning graphs by changing the corresponding treatments and then exploit the counterfactual outcomes in a contrastive learning framework. We conduct extensive experiments on three real-world datasets, and the experimental results demonstrate the superiority of our proposed Graph CA method compared with several state-of-the-art baselines.展开更多
Programming ability has become one of the most practical basic skills,and it is also the foundation of software development.However,in the daily training experiment,it is difficult for students to find suitable exerci...Programming ability has become one of the most practical basic skills,and it is also the foundation of software development.However,in the daily training experiment,it is difficult for students to find suitable exercises from a large number of topics provided by numerous online judge(OJ)systems.Recommending high passing rate topics with an effective prediction algorithm can effectively solve the problem.Directly applying some common prediction algorithms based on knowledge tracing could bring some problems,such as the lack of the relationship among programming exercises and dimension disaster of input data.In this paper,those problems were analyzed,and a new prediction algorithm was proposed.Additional information,which represented the relationship between exercises,was added in the input data.And the input vector was also compressed to solve the problem of dimension disaster.The experimental results show that deep knowledge tracing(DKT)with side information and compression(SC)model has an area under the curve(AUC)of 0.7761,which is better than other models based on knowledge tracing and runs faster.展开更多
Knowledge tracing(KT),aiming at mining students’mastery of knowledge by their exercise records and predicting their performance on future test questions,is a critical task in educational assessment.While researchers ...Knowledge tracing(KT),aiming at mining students’mastery of knowledge by their exercise records and predicting their performance on future test questions,is a critical task in educational assessment.While researchers achieve tremendous success with the rapid development of deep learning techniques,current KT tasks fall into the cracks from real-world teaching scenarios.Relying on extensive student data heavily and predicting numerical performances solely differ from the settings where teachers assess students’knowledge state from limited practices and provide explanatory feedback.To fill this gap,this study explores a new task formulation,namely,explainable few-shot KT.By leveraging the powerful reasoning and generation abilities of large language models(LLMs),this study then proposes a cognition-guided framework that can track students’knowledge from a few students’records while providing natural language explanations.Experimental results from three widely used datasets show that LLMs can perform comparable or superior to competitive deep KT methods.Finally,this study discusses potential directions and calls for future improvements.展开更多
Deep learning has found widespread application across diverse domains owing to its exceptional performance.Nevertheless,the lack of transparency in deep learning models’decision-making processes undermines their usab...Deep learning has found widespread application across diverse domains owing to its exceptional performance.Nevertheless,the lack of transparency in deep learning models’decision-making processes undermines their usability,especially in critical contexts.While researchers have made noteworthy advancements in explaining these models,they have frequently overlooked the differences between static and temporal models during explanation generation.In temporal models,features change over time,posing new challenges in the generation of explanations.Though extensive research has been dedicated to surmounting these hurdles,a survey summarizing these contributions is currently absent.To bridge this gap,this paper endeavors to summarize existing methods and their contributions in terms of both static and temporal models,highlighting their disparities.Additionally,we propose an innovative classification approach based on the comprehensibility of explanations,demonstrating that different explanation methods vary in their understandability for users.Finally,to assess the limitations of the explanation capabilities of existing methods,we specifically choose knowledge tracing to analyze the evolution of explanation methods in this context of temporal modeling and interpretations.展开更多
In educational practice,teachers often need to manually assemble an exercise collection as a class quiz or a homework assignment.A well-assembled exercise collection needs to have the proper difficulty index and discr...In educational practice,teachers often need to manually assemble an exercise collection as a class quiz or a homework assignment.A well-assembled exercise collection needs to have the proper difficulty index and discrimination index so that it can better develop students'abilities.In this paper,we propose an exercise collection auto-assembling framework,in which a teacher provides the target values of difficulty and discrimination indices and a qualified exercise collection is automatically assembled.The framework consists of two stages.At the answer prediction stage,a knowledge tracing model is utilized to predict the students'answers to unseen exercises based on their history interaction records.In addition,to better represent the exercises in the model,we propose exercise embeddings and design a pre-training approach.At the collection assembling stage,we propose a deep reinforcement learning model to assemble the required exercise collection effectively.Since the knowledge tracing model in the first stage has different confidences in the predicted answers,it is also taken into account in the objective.Experimental results show the effectiveness and efficiency of the proposed framework.展开更多
Knowledge tracing aims to track students’knowledge status over time to predict students’future performance accurately.In a real environment,teachers expect knowledge tracing models to provide the interpretable resul...Knowledge tracing aims to track students’knowledge status over time to predict students’future performance accurately.In a real environment,teachers expect knowledge tracing models to provide the interpretable result of knowledge status.Markov chain-based knowledge tracing(MCKT)models,such as Bayesian Knowledge Tracing,can track knowledge concept mastery probability over time.However,as the number of tracked knowledge concepts increases,the time complexity of MCKT predicting student performance increases exponentially(also called explaining away problem).When the number of tracked knowledge concepts is large,we cannot utilize MCKT to track knowledge concept mastery probability over time.In addition,the existing MCKT models only consider the relationship between students’knowledge status and problems when modeling students’responses but ignore the relationship between knowledge concepts in the same problem.To address these challenges,we propose an inTerpretable pRobAbilistiC gEnerative moDel(TRACED),which can track students’numerous knowledge concepts mastery probabilities over time.To solve explain away problem,we design long and short-term memory(LSTM)-based networks to approximate the posterior distribution,predict students’future performance,and propose a heuristic algorithm to train LSTMs and probabilistic graphical model jointly.To better model students’exercise responses,we proposed a logarithmic linear model with three interactive strategies,which models students’exercise responses by considering the relationship among students’knowledge status,knowledge concept,and problems.We conduct experiments with four real-world datasets in three knowledge-driven tasks.The experimental results show that TRACED outperforms existing knowledge tracing methods in predicting students’future performance and can learn the relationship among students,knowledge concepts,and problems from students’exercise sequences.We also conduct several case studies.The case studies show that TRACED exhibits excellent interpretability and thus has the potential for personalized automatic feedback in the real-world educational environment.展开更多
Personalized education,tailored to individual stu-dent needs,leverages educational technology and artificial intelligence(AI)in the digital age to enhance learning ef-fectiveness.The integration of AI in educational p...Personalized education,tailored to individual stu-dent needs,leverages educational technology and artificial intelligence(AI)in the digital age to enhance learning ef-fectiveness.The integration of AI in educational platforms provides insights into academic performance,learning pref-erences,and behaviors,optimizing the personal learning process.Driven by data mining techniques,it not only ben-efits students but also provides educators and institutions with tools to craft customized learning experiences.To offer a comprehensive review of recent advancements in person-alized educational data mining,this paper focuses on four primary scenarios:educational recommendation,cogni-tive diagnosis,knowledge tracing,and learning analysis.This paper presents a structured taxonomy for each area,compiles commonly used datasets,and identifies future re-search directions,emphasizing the role of data mining in enhancing personalized education and paving the way for future exploration and innovation.展开更多
基金supported by National Natural Science Foundation of China(Nos.62266054 and 62166050)Key Program of Fundamental Research Project of Yunnan Science and Technology Plan(No.202201AS070021)+2 种基金Yunnan Fundamental Research Projects(No.202401AT070122)Yunnan International Joint Research and Development Center of China-Laos-Thailand Educational Digitalization(No.202203AP140006)Scientific Research Foundation of Yunnan Provincial Department of Education(No.2024Y159).
文摘In the field of intelligent education,the integration of artificial intelligence,especially deep learning technologies,has garnered significant attention.Knowledge tracing(KT)plays a pivotal role in this field by predicting students’future performance through the analysis of historical interaction data,thereby assisting educators in evaluating knowledgemastery and tailoring instructional strategies.Traditional knowledge tracingmethods,largely based on Recurrent Neural Networks(RNNs)and Transformer models,primarily focus on capturing long-term interaction patterns in sequential data.However,these models may neglect crucial short-term dynamics and other relevant features.This paper introduces a novel approach to knowledge tracing by leveraging a pure Multilayer Perceptron(MLP)architecture.We proposeMixerKT,a knowledge tracing model based on theHyperMixer framework,which uniquely integrates global and localMixer feature extractors.This architecture enables more effective extraction of both long-terminteraction trends and recent learning behaviors,addressing limitations in currentmodels thatmay overlook these key aspects.Empirical evaluations on twowidely-used datasets,ASSIS Tments2009 and Algebra2005,demonstrate that MixerKT consistently outperforms several state-of-the-art models,including DKT,SAKT,and Separated Self-Attentive Neural Knowledge Tracing(SAINT).Specifically,MixerKT achieves higher prediction accuracy,highlighting its effectiveness in capturing the nuances of learners’knowledge states.These results indicate that our model provides a more comprehensive representation of student learning patterns,enhancing the ability to predict future performance with greater precision.
文摘Knowledge tracing is the key component in online individualized learning,which is capable of assessing the users'mastery of skills and predicting the probability that the users can solve specific problems.Available knowledge tracing models have the problem that the assessments are not directly used in the predictions.To make full use of the assessments during predictions,a novel model,named deep knowledge tracing embedding neural network(DKTENN),is proposed in this work.DKTENN is a synthesis of deep knowledge tracing(DKT)and knowledge graph embedding(KGE).DKT utilizes sophisticated long short-term memory(LSTM)to assess the users and track the mastery of skills according to the users'interaction sequences with skill-level tags,and KGE is applied to predict the probability on the basis of both the embedded problems and DKT's assessments.DKTENN outperforms performance factors analysis and the other knowledge tracing models based on deep learning in the experiments.
基金supported by the Natural Science Foundation of China (62372277)the Natural Science Foundation of Shandong Province (ZR2022MF257, ZR2022MF295)Humanities and Social Sciences Fund of the Ministry of Education (21YJC630157)。
文摘With the popularity of online learning in educational settings, knowledge tracing(KT) plays an increasingly significant role. The task of KT is to help students learn more effectively by predicting their next mastery of knowledge based on their historical exercise sequences. Nowadays, many related works have emerged in this field, such as Bayesian knowledge tracing and deep knowledge tracing methods. Despite the progress that has been made in KT, existing techniques still have the following limitations: 1) Previous studies address KT by only exploring the observational sparsity data distribution, and the counterfactual data distribution has been largely ignored. 2) Current works designed for KT only consider either the entity relationships between questions and concepts, or the relations between two concepts, and none of them investigates the relations among students, questions, and concepts, simultaneously, leading to inaccurate student modeling. To address the above limitations,we propose a graph counterfactual augmentation method for knowledge tracing. Concretely, to consider the multiple relationships among different entities, we first uniform students, questions, and concepts in graphs, and then leverage a heterogeneous graph convolutional network to conduct representation learning.To model the counterfactual world, we conduct counterfactual transformations on students’ learning graphs by changing the corresponding treatments and then exploit the counterfactual outcomes in a contrastive learning framework. We conduct extensive experiments on three real-world datasets, and the experimental results demonstrate the superiority of our proposed Graph CA method compared with several state-of-the-art baselines.
文摘Programming ability has become one of the most practical basic skills,and it is also the foundation of software development.However,in the daily training experiment,it is difficult for students to find suitable exercises from a large number of topics provided by numerous online judge(OJ)systems.Recommending high passing rate topics with an effective prediction algorithm can effectively solve the problem.Directly applying some common prediction algorithms based on knowledge tracing could bring some problems,such as the lack of the relationship among programming exercises and dimension disaster of input data.In this paper,those problems were analyzed,and a new prediction algorithm was proposed.Additional information,which represented the relationship between exercises,was added in the input data.And the input vector was also compressed to solve the problem of dimension disaster.The experimental results show that deep knowledge tracing(DKT)with side information and compression(SC)model has an area under the curve(AUC)of 0.7761,which is better than other models based on knowledge tracing and runs faster.
基金supported by the National Natural Science Foundation of China(Grant No.62377002)the SMP-Zhipu.AI Large Model Cross-Disciplinary Fund(Grant No.20240211).
文摘Knowledge tracing(KT),aiming at mining students’mastery of knowledge by their exercise records and predicting their performance on future test questions,is a critical task in educational assessment.While researchers achieve tremendous success with the rapid development of deep learning techniques,current KT tasks fall into the cracks from real-world teaching scenarios.Relying on extensive student data heavily and predicting numerical performances solely differ from the settings where teachers assess students’knowledge state from limited practices and provide explanatory feedback.To fill this gap,this study explores a new task formulation,namely,explainable few-shot KT.By leveraging the powerful reasoning and generation abilities of large language models(LLMs),this study then proposes a cognition-guided framework that can track students’knowledge from a few students’records while providing natural language explanations.Experimental results from three widely used datasets show that LLMs can perform comparable or superior to competitive deep KT methods.Finally,this study discusses potential directions and calls for future improvements.
基金supported by the National Natural Science Foundation of China under Grant Nos.62272093,62137001,and 62372097.
文摘Deep learning has found widespread application across diverse domains owing to its exceptional performance.Nevertheless,the lack of transparency in deep learning models’decision-making processes undermines their usability,especially in critical contexts.While researchers have made noteworthy advancements in explaining these models,they have frequently overlooked the differences between static and temporal models during explanation generation.In temporal models,features change over time,posing new challenges in the generation of explanations.Though extensive research has been dedicated to surmounting these hurdles,a survey summarizing these contributions is currently absent.To bridge this gap,this paper endeavors to summarize existing methods and their contributions in terms of both static and temporal models,highlighting their disparities.Additionally,we propose an innovative classification approach based on the comprehensibility of explanations,demonstrating that different explanation methods vary in their understandability for users.Finally,to assess the limitations of the explanation capabilities of existing methods,we specifically choose knowledge tracing to analyze the evolution of explanation methods in this context of temporal modeling and interpretations.
基金This work was supported by the National Natural Science Foundation of China under Grant Nos.62072261 and 61925205,and Huawei and TAL education.
文摘In educational practice,teachers often need to manually assemble an exercise collection as a class quiz or a homework assignment.A well-assembled exercise collection needs to have the proper difficulty index and discrimination index so that it can better develop students'abilities.In this paper,we propose an exercise collection auto-assembling framework,in which a teacher provides the target values of difficulty and discrimination indices and a qualified exercise collection is automatically assembled.The framework consists of two stages.At the answer prediction stage,a knowledge tracing model is utilized to predict the students'answers to unseen exercises based on their history interaction records.In addition,to better represent the exercises in the model,we propose exercise embeddings and design a pre-training approach.At the collection assembling stage,we propose a deep reinforcement learning model to assemble the required exercise collection effectively.Since the knowledge tracing model in the first stage has different confidences in the predicted answers,it is also taken into account in the objective.Experimental results show the effectiveness and efficiency of the proposed framework.
基金supported by the National Natural Science Foundation of China(Grant Nos.62272093,62137001,U1811261,and 61902055).
文摘Knowledge tracing aims to track students’knowledge status over time to predict students’future performance accurately.In a real environment,teachers expect knowledge tracing models to provide the interpretable result of knowledge status.Markov chain-based knowledge tracing(MCKT)models,such as Bayesian Knowledge Tracing,can track knowledge concept mastery probability over time.However,as the number of tracked knowledge concepts increases,the time complexity of MCKT predicting student performance increases exponentially(also called explaining away problem).When the number of tracked knowledge concepts is large,we cannot utilize MCKT to track knowledge concept mastery probability over time.In addition,the existing MCKT models only consider the relationship between students’knowledge status and problems when modeling students’responses but ignore the relationship between knowledge concepts in the same problem.To address these challenges,we propose an inTerpretable pRobAbilistiC gEnerative moDel(TRACED),which can track students’numerous knowledge concepts mastery probabilities over time.To solve explain away problem,we design long and short-term memory(LSTM)-based networks to approximate the posterior distribution,predict students’future performance,and propose a heuristic algorithm to train LSTMs and probabilistic graphical model jointly.To better model students’exercise responses,we proposed a logarithmic linear model with three interactive strategies,which models students’exercise responses by considering the relationship among students’knowledge status,knowledge concept,and problems.We conduct experiments with four real-world datasets in three knowledge-driven tasks.The experimental results show that TRACED outperforms existing knowledge tracing methods in predicting students’future performance and can learn the relationship among students,knowledge concepts,and problems from students’exercise sequences.We also conduct several case studies.The case studies show that TRACED exhibits excellent interpretability and thus has the potential for personalized automatic feedback in the real-world educational environment.
基金supported by the National Natural Science Foundation of China(No.62377002).
文摘Personalized education,tailored to individual stu-dent needs,leverages educational technology and artificial intelligence(AI)in the digital age to enhance learning ef-fectiveness.The integration of AI in educational platforms provides insights into academic performance,learning pref-erences,and behaviors,optimizing the personal learning process.Driven by data mining techniques,it not only ben-efits students but also provides educators and institutions with tools to craft customized learning experiences.To offer a comprehensive review of recent advancements in person-alized educational data mining,this paper focuses on four primary scenarios:educational recommendation,cogni-tive diagnosis,knowledge tracing,and learning analysis.This paper presents a structured taxonomy for each area,compiles commonly used datasets,and identifies future re-search directions,emphasizing the role of data mining in enhancing personalized education and paving the way for future exploration and innovation.
