Research On Learning Recommendation System Via Student Achievement Analysis

With the rapid development of the Internet and information technology,people have gradually entered the era of information overload from the era of information scarcity.In the field of education,with the rapid growth of the number of teaching resources,the traditional campus teaching model can no longer meet the needs of students.Therefore,personalized learning recommendations came into being,bringing new ideas to the construction of educational information.Personalized recommendation learning with learners as the main body has become a development trend and research hotspot of online teaching.How to extract the learning resources that the learners are interested in or need from the massive data information,and recommend the corresponding learning resources according to the characteristics of the learners,is of great significance for improving the learning efficiency and effect of the learners.Most of the current learning recommendation applications use cognitive diagnosis method(DINA),collaborative filtering method(CF),and probability matrix decomposition method(PMF).However,the cognitive diagnosis method can not probabilize the knowledge level,and it is easy to cause data loss in the application process.The collaborative filtering method and the probability matrix decomposition method do not analyze the mastery of the student knowledge points.In addition,all methods mentioned above are not considered the effects of time decay factors.In this paper,a recommendation method based on performance analysis is proposed for the problem of poor recommendation of personalized learning recommendation algorithm.The main work contents and innovations are as follows:(1)For a large number of student test data,the reliability,validity,difficulty and discrimination of classical test theory are used to analyze the quality of each test paper to confirm whether the test paper and related data have the value of further research.Through analysis,data that does not meet the conditions are deleted,and high-quality data information is collated for further research.(2)In order to solve the problems existing in the existing model,a new probability model of student knowledge points is proposed.Considering the influence of time importance,student performance and time,the three important factors of knowledge point weight,loss rate and time decay factor are extracted to construct the model,and the knowledge point recommendation list is obtained through the model.(3)In the process of learning resource recommendation,the performance rating is based on the student's performance.Then according to the student's grade,the TOP-N recommendation strategy is used to recommend the learning resources of the corresponding difficulty level to cover the weak knowledge points.Finally,the personalized learning recommendation system is implemented by using the proposed strategy,which achieves the effect of leak detection and defect filling for students.