Research On Students’ Mastery Of Knowledge Points Based On Knowledge Tracking

In recent years,with the rapid development and popularization of computer networks,online education platforms have experienced a spurt of development in the commercial field.In this process,a large amount of learner response data is generated.These data contain a large amount of information about the degree of knowledge mastery during the learning behavior of the learner.Through analysis and research of this information,the educator can not only understand the learning progress of each learner,but also can recommend efficient learning paths and appropriate learning resources to learners,so as to teach students in accordance with their aptitude.However,since traditional educational data mining methods mostly focus on the question of whether students answer correctly,there are limitations in enlightening educators.Therefore,this article uses the knowledge tracking method to deeply dig into the students’ knowledge mastery and improve the knowledge tracking model,which is very necessary in today’s online education business.Based on the student response data from the Sudoku module of the online teaching platform,this paper first constructs four new features:maximum number of attempts,total number of attempts,average response time and correctness rate based on the original data,and uses the support vector machine model to predict the correctness of the first answer of the student’s special test.The accuracy rate,precision rate,recall rate,and F1 value of the model are 0.64,0.81,0.62,0.61,respectively.The model performance is not ideal.It can be considered that the support vector machine model has no predictive significance on this data set and has not learned to effective information.Then,in the scene of two knowledge points,this paper establishes the project response theory model and the Bayesian knowledge tracking model based on the sequence data of the students’ answering state in turn,and evaluates the students’performance in the special evaluation through classroom exercises and homework data.The correct rate of the first answer is predicted,and the conclusion is reached:the AUC values obtained by the two models on the two assessment questions are 0.732,0.749 and 0.750,0.771 respectively.Comparing the effects of the three models,it is found that the Bayesian knowledge tracking model has the best accuracy in predicting the first answer state of the test questions,but the effects of the three models are not good. Finally,based on the answering status of each grid,the students’ mastery of knowledge points is divided into three levels:completely unmastered,partially mastered,and fully mastered,and an improved Bayesian knowledge point diagnostic model is established to predict the degree of student knowledge.The Kappa coefficient of the model is 0.778,and the predicted result of the model is highly consistent with the actual classification.It can be concluded that the improved Bayesian knowledge tracking model based on multi-classification performs best on this data set.