Research On Personalized Exercise Recommendation Algorithm Based On Education Data Mining

With the continuous development of the information age,especially the steady advancement of the mobile Internet,people have quickly entered the era of big data.In contrast to the rapid growth in information volume,people have not seen significant changes in the speed of learning new knowledge and acquiring new capabilities.The personalized recommendation of the information is that the user only needs to process the recommended local information,and the effect of the processing of all the information can be achieved.In the process of people's learning,due to various reasons,people's mastery of knowledge will be differentiated.In the process of teaching,teachers have limited energy,and their lectures and after-school assignments can only take care of most people.Therefore,combined with the data mining method,personalized teaching assistance for different students has become a hot research topic.Education Data Mining(EDM)is an important research direction for personalized teaching assistance.In recent years,the increase of online video teaching,especially the rise of online real-time classrooms,including MOOC,Coursera,Khan Academy,School Online,etc.,has rapidly accumulated a large amount of pure electronic data.These data provide a wealth of material for educational data mining research.As an important part of the education process,the recommendation of personalized exercises is of great significance.According to the individualized situation of different students,the specific exercises recommendation can effectively improve the teaching quality.However,the recommendation of personalized exercises is still facing enormous challenges.First of all,how to accurately obtain the students' knowledge of the overall learning and the knowledge points that have not yet been mastered according to the students' historical learning behaviors,so as to accurately model the students,this still has a huge room for improvement.Secondly,how to carry out reasonable electronic modeling of the knowledge points that students need to master,and based on the student's cognitive level model,conduct reasonable individualized exercises recommendation,so that students can find their disadvantage more quickly and accurately.This is also one of the issues that researchers focus on.This article mainly does the following work:1.On the basis of summarizing the field of educational data mining,especially the previous research results of personalized problem recommendation,the author first analyzes and summarizes the shortcomings of existing achievements and possible improvement strategies.2.In view of the existing student cognitive level model,the problem that the user modeling can not truly and effectively reflect the shortcomings of the students' true cognitive state,the author proposes a new method of student cognitive level modeling.In the new model,students' cognitive modeling is divided into comprehensive cognitive level modeling and specific knowledge point cognitive level modeling.The two modeling methods are constructed from different angles,and finally combined to reflect the students' cognitive level and can get better results.3.For the traditional exercises recommendation algorithm,only consider the difficulty of the exercises or the shortcomings of the knowledge point,combining with the above user's cognitive level,this paper proposes a comprehensive exercises recommended algorithm(NPERA).This algorithm guarantees the full coverage of the knowledge points with poor mastery.At the same time,it recommends the exercises suitable for the students' level of difficulty,which can improve the students' experience of the exercises while ensuring the effect of the exercises.4.The algorithm proposed in this paper is verified by experiments on the real dataset,and it shows that the algorithm can have advantages in many indicators.