Construction And Implementation Of A Personalized Recommendation Model Of Learning Resources Based On Deep Neural Network

Personalized learning is an important goal,direction and trend of educational development,and has been highly valued by the state,the Ministry of Education,and schools at all levels.With rapid development of the Internet age,numerous sharable and reusable learning resources have brought us convenience and challenges along the way.The personalized recommendation of learning resources can promote the development of personalized learning and solve the problem of information disorientation.Artificial intelligence is widely used in e-commerce,medical care,transportation and other fields,and has achieved fruitful results.Its further application in the field of education will undoubtedly bring new opportunities and vitality to personalized recommendation of learning resources,which is going to be the focus of educational research.Through in-depth research in this field,the study found that there are abundant researches related to personalized recommendation of learning resources,while research based on deep learning starts late and is relatively limited.By first analyzing the main content,basic requirements,as well as deficiencies of personalized recommendation of learning resources,and the basic principles,main models and application fields of deep learning,this study summarizes the characteristics of personalized recommendation of learning resources and puts forward a personalized recommendation model based on deep neural network under the guidance of personalized learning theory and connectivist learning theory.The model optimizes its efficiency,diversity and timeliness under the support of deep learning technology.Then,for the modules of resource filtering,resource recommendation and resource display,it designed a similarity-ranking-based resource filtering algorithm,a deep neural network-based resource recommendation and a resource display algorithm,collected and processed the data from online learning platform through Python language programming and other data processing technology.This model is pre-trained through the processed data set,and the algorithm designed in this research is compared to the classic algorithm to verify its rationality and effectiveness.Finally,taking We Chat official account as an example,a personalized recommendation platform for learning resources based on deep neural network is designed.The platform visualizes the proposed model,and the platform is implemented by using Django framework and open source interfaces and libraries.With the help of Intranet penetration tools,the platform is released online to some learners for trial,and then evaluation remarks are collected through questionnaire survey and interview.Results show that users have a good evaluation of accuracy,diversity and timeliness of the resources recommended by the platform,but the ranking evaluation of the resource list needs to be improved.So a conclusion is drawn that the ranking of the recommended learning resource list cannot solely rely on prediction ratings.It is hoped that this research will provide reference for personalized recommendation of learning resources based on deep learning technology.