Research On Educational Resource Recommendation Technology Based On Learner Model

The vigorous development of educational resources information has caused the problems of educational resources "information overload" and learners "knowledge loss".In order to alleviate this problem,many researchers have introduced recommendation technology to achieve personalized recommendation of educational resources.In response to the above problems,researchers have been inspired by e-commerce recommendation systems and movie recommendation systems,and applied collaborative filtering and other recommendation methods to education resource.Due to the particularity of educational resources,the general recommendation method has the problem of low performance in educational resources recommendation,which leads to poor user experience.At the same time,new users' cold start,recommendation efficiency,and low accuracy still exist objectively.In this paper,combined with deep learning and natural language processing technology,a hybrid recommendation algorithm for educational resources based on the learner model is proposed for the field of educational resource recommendation.The proposed method not only studies the basic personal information and behavior information of the target user,but also Calculation of resources,learning ability and level of other users.The experimental data set is the real data collected by a web crawler to an educational website.The main work includes:1.Alleviate the cold start problem.Combining deep learning and natural language processing technology,firstly,use natural language processing technology which under the framework of deep learning to analyze the registration information and personal information of new users are for classification.Secondly,according to the result of classification to find out the corresponding category of target users.Then we calculate the learning similarity between the target user and other users in the category to find out similar user set.Finally use the fusion calculation to determine the end of similar user set,and make a personalized educational resources recommendation.2.Improve computing efficiency.First,we classify the target users,calculate the similarity within the class,and then calculate the similarity of learning ability between the target users and other users of the same class.Intra class computing reduces the computing time,and learning ability computing can further find out the users who match the learning ability and interest of the target users,and improve the accuracy of recommendation.3.Improve the accuracy of recommendation from multiple perspectives.In the recommendation method,the penalty factor of popular resources and the dynamic change factor of user interest are introduced.To solve the dynamic change of user interest and the influence of popular educational resources on the recommendation results.The experimental results show that,in comparison with the traditional recommendation method,the proposed method can effectively alleviate the problem of cold start,improve the calculation efficiency and the accuracy of the recommendation result.This method has achieved good results in the recommendation of educational resources and has certain Practical significance.