Research On Personalized Learning Model And Application Based On Online Course

In the era of "Internet + education",with the rapid development of data analysis and information science,online courses have been widely promoted throughout the whole education cycle.Learners are no longer restricted by time and space,and can be granted high-quality education anytime and anywhere,which fully expands learners’ learning form and education method.However,due to the differences in learning habits and background knowledge among learners,and the lack of communication between teachers and students,it is difficult to realize teachers’ teaching in accordance with their aptitude and learners’ personalized development,which affects learners’ learning effect to a great extent.In addition,the increasing scale of learners has also exacerbated this situation.Because the learner-centered personalized learning model can reflect the personalized differences among learners in multiple dimensions,it has become one of the feasible methods to change the development status of online courses.At present,taking online courses as the application scenario,most personalized learning models generally pay attention to the learning behavior characteristics of learners.They conduct personalized learning diagnoses for learners by analyzing the behavior characteristics,ignoring the mining of the causal relationship between learning behavior and learning performance,as well as the personalized discovery of the importance of knowledge points in the knowledge state.In view of the above shortcomings,this thesis aims to provide strong support for the research of the personalized learning model and decision support for the development of online courses,by mining the causal effect between learning behavior and learning performance and the personalized importance of knowledge points.The main research work of this thesis is as follows:1)According to the learning interaction data in the curriculum forum,the causal effects of learning interaction and the intensity of different interaction categories on learning performance are evaluated through "counterfactual" causal inference,which alleviates the influence of the endogenous and sample selection errors of the regression model on the results,and reconciles the differences in the existing research to a certain extent.2)A personalized knowledge map with multiple types of nodes is adaptively generated from the learning-evaluation data.The map can not only link learners and knowledge-points into a unified map,but also capture the structure between knowledge-points and learners’ mastery of knowledgepoints.On this basis,a ranking method based on a random walk is proposed,which is used to calculate the importance of each learner’s knowledge-points.When walking,learners’ differences in mastery and understanding of knowledge-points and the difficulty of knowledge-points themselves can be considered at the same time.According to the causal effects of learning interaction and learning performance,the author conducted a large number of experiments on a real data set.The experimental results show that the knowledge-point importance discovery method proposed in this thesis has advantages compared with other relevant baselines,which proves the effectiveness of the learner component and difficulty level component in the model.3)Based on the above research results,a personalized learning prototype system is designed and developed from the level of practical application.According to the online behavior characteristics and learning-evaluation data of learners,the system realizes the two kinds of research in this thesis,that is,the inference of causal effects between behaviors and the calculation of the personalized importance of knowledge points,and realizes the cluster analysis of the importance of knowledge points.The system visualizes the relevant results and provides decision support for teachers.And the system can accurately analyze the knowledge state of learners so that learners can get personalized development in online courses.