Research On Personalized Exercise Recommendation Algorithm Based On Online Learning Behavior Analysis

Nowadays,the rapid development of Internet technology has promoted the continuous in-depth integration and development of the education field.Online education has ushered in a period of rapid development.Therefore,online learning platforms have been widely used.For learners,the online learning platform has the advantages of rich learning content and easy access to knowledge points.This new learning method has brought new learning experience to learners,but online learning is also facing the problems of "data flooding" and "knowledge Trek".In order to solve the above problems,personalized learning recommendation research has become the focus of educational researchers.As a branch of personalized learning recommendation,personalized exercise recommendation is of great significance to improve the quality of online education and realize intelligent learning.In order to realize learners’ personalized exercise recommendation,this paper takes the online evaluation system as the research object.As an online learning auxiliary platform,the online evaluation system can record the online learning behavior information such as learners’ sign in information,the number of exercises completed and the number of exercises submitted,cluster according to the characteristics of learners’ online learning behavior,and recommend appropriate exercises to learners in combination with collaborative filtering recommendation technology,So as to help learners improve learning efficiency and learning experience.The specific research work of this paper is as follows:(1)This paper proposes a collaborative filtering recommendation algorithm based on improved learner similarity computation.Firstly,the improved algorithm takes into account the influence of the number of questions that learners do together,the numerical difference of the number of errors in questions and the difficulty of exercises on the similarity calculation of learners.Three corresponding correction factors are introduced to improve the traditional similarity calculation method.Secondly,the improved algorithm calculates the similarity between learners by introducing two balance factors to improve the similarity calculation method.Finally,the experimental results show that the similarity calculation method based on the fusion of correction factor and weight factor can calculate the similarity between learners more accurately.(2)This paper proposes a collaborative filtering recommendation(ucis-cf)algorithm based on user clustering and similarity.Ucis-cf algorithm combines clustering technology and improved collaborative filtering recommendation technology to realize learners’ personalized exercise recommendation.The introduction of clustering technology effectively improves the efficiency of nearest neighbor selection of collaborative filtering recommendation algorithm and reduces the scale of similarity calculation of collaborative filtering recommendation algorithm.The improved collaborative filtering recommendation algorithm optimizes the similarity calculation between learners and improves the recommendation performance of the algorithm to a certain extent.The comparative experimental results show that the optimized ucis-cf algorithm has better Mae value,accuracy,recall and F1 value than other collaborative filtering recommendation algorithms based on traditional similarity calculation,and has better reliability and stability.