Research On MOOC Recommendation Methods For Combining User Behavior And Courses Attribute Data

MOOC recommendation methods can effectively solve the selection problem of the massive courses in the current Internet.But the mainstream MOOC recommendation methods are mostly based on explicit information such as user tags or test results.This kind of method not only has poor user interface friendliness,but also does not fully consider the implicit information from the interactive information between users and MOOC websites,resulting in low accuracy and interpretation of MOOC recommendation results.Therefore,it is of theoretical and practical significance to study how to obtain the implicit preference features from the interactive information between users and MOOC websites,and build a more accurate and more interpretative MOOC recommendation method.In view of the problem that the results of the current MOOC recommendation methods are less interpretative,this paper analyzed the available data from the MOOC website.We can extract features that can describe attributes of courses from course related text data,so we can apply course features to MOOC recommendation based on the that the course features can effectively alleviate the problem of poor interpretation that caused by the current MOOC recommendation methods only use user rating to recommend.In view of the problem that the current MOOC recommendation methods use traditional similarity calculation methods to calculate the similarity of courses or users,which leads to the low accuracy of recommendation results,so this paper proposed a new similarity calculation method to calculate the similarity of courses.Three workflows of the MOOC recommendation that based on the general workflow of recommendation are given,which are course feature extraction,user preference prediction and recommendation result generation.Based on user behavior and course features,a more accurate and more interpretative method of MOOC recommendation is designed.First,we selected the LDA to extract course features,then constructed the MOOC user behavior matrix.Finally,we designed a new MOOC recommendation methods for combining user behavior and courses related data based on GLSLIM model.Through the comparison experimental between the MOOC recommendation method in this paper and other MOOC recommendation methods,it is proved that the recommendation method in this paper has a good effect in all indicators.At the same time,the influence of feature dimension and the number of user cluster on the recommendation effect of MOOC is discussed.