Research On Knowledge Point Recommendation Algorithm Based On Learning Behavior Analysis

Due to the complexity and diversity of the structure of knowledge points and the design of the teaching process,online courses lead to different learning methods of knowledge points for learners with the same cognitive ability,which lead to differences in learning effects.We study from two angles: similarity calculation between learners and learning pattern mining of knowledge point sequence by learners,and propose two novel knowledge recommendation algorithms.The main research contents are as follows:(1)To solve the problem of inaccurate description of similarity in the classical collaborative filtering recommendation algorithm,we propose a knowledge recommendation algorithm based on similarity optimization.The algorithm fully considers the difference of learners' relative difficulty coefficient of knowledge points and evaluation criteria,and integrates the knowledge point information learned by non-associated learners to optimize the similarity calculation between learners.Compared with the classic recommendation algorithms(Cos ? Pearson ? ICF and Jac RA algorithm),the proposed algorithm has better advantages in the performance evaluation indexes of prediction accuracy,precision rate,recall rate,and F1 value.(2)In view of the limitation of the classical Pearson correlation coefficient in similarity calculation and the failure of classical learning resource recommendation algorithms to mine learners' learning behavior characteristics effectively,we propose a knowledge point recommendation algorithm based on enhanced correction factor and weighted sequential pattern mining(ECF-WSPM).The algorithm divides learners into different groups according to their cognitive levels,constructs an enhanced correction factor through the local and global differences of the relative difficulty coefficient of knowledge points,and reconstructs the similarity calculation model between learners.We combine the conceptual interaction achievement degree of knowledge points with learners' sequential learning patterns to effectively mine the characteristics of learners' learning behavior,so as to generate the corresponding recommended list of knowledge points for the target learners.Compared with the learning resource recommendation algorithms(GSP-CA-CF and UCF-SPM algorithm),our proposed algorithm improves the performance evaluation index of the algorithm in terms of prediction accuracy,precision rate,recall rate,and F1 value.The effectiveness of the proposed algorithm is verified by the data set obtained from the web-based learning platform developed by our university,the experimental results show that good results have been achieved.