Research And Implementation Of Adaptive Learning System Based On Deep Learning

With the rapid development of information technology,distance education has become possible.Adaptive learning system,which can overcome the problem of educational resources' maldistribution and model of teaching's singleness in traditional China's traditional education,could dynamically adjust the presentation of learning content according to the learner's learning situation.And it has great significance for improving teaching quality and achieving efficient learning.However,compared with the foreign countries which already have mature adaptive learning system,such as Moodle learning management system and Knewton personalized learning platform,the domestic research on adaptive learning system is still in the theoretical stage.And a few of the learning systems which have been promoted and used are generally lack of intelligence.The research and development of adaptive learning system is one of the hotspots in educational field at this stage.Personalized learning resource recommendation is the main way to realize the adaptability of learning system.Data mining is used to acquire students' learning behavior data.The learner model is constructed through learning analysis technology and then provide customized learning resources.This method helps Students get rid of the dilemma of "information explosion".Firstly,through consulting a large number of documents,it is found that the traditional learning resource recommendation method mainly stays at the level of questions.It ignores the essence of learning is to learn the knowledge points and students' learning process is influenced by their learning log which is temporal.In recent years,with the rapid development of artificial intelligence,combining recommended methods with deep learning neural network is an important direction.The Bi-LSTM is good at processing timing problems and considers log to recommend contents.The Deep Autoencoder is good at processing complex information and has a good performance in predicting scores in the collaborative recommendation field.Then,based on the above situation,this paper proposes a personalized learning resource recommendation method based on Bi-LSTM and Deep Autoencoder.Firstly,the two-step collaborative filtering exercise recommendation based on the knowledge point is carried out.Then predict the correct rate of the target student on the recommended question and the score on the key step of the recommended question.And the final result synergistically decided by the above two prediction results.After comparing with the traditional recommendation algorithms,we can find that the personalized learning resource recommendation method proposed in this paper has the better stability and interpretability.Finally,according to the functional requirements and structural characteristics of the adaptive learning system,we design and develop an adaptive learning system and integrate the personalized learning resource recommendation algorithm proposed in this paper into it.The adaptive learning system which generates learning resource and overcomes the dilemma of "information explosion" has a good performance in achieving personalized learning.