Research On Personalized Online Learning Resource Recommendation Based On Deep Learning

With the rapid rise of new computer technology and the increase of the number of online learning platforms,the forms of learning resources are becoming more and more diversified.At the same time,the scale of large educational data is exponentially increasing,which makes it more difficult for learners to choose appropriate resources from the vast amount of learning resources for learning.How to screen valuable information from the vast amount of data has become an online learning leader.A big problem in domain.For online learning platform,the introduction of personalized recommendation technology can effectively alleviate this problem.Recommending personalized learning resources to learners can effectively help learners deal with resource overload and knowledge fragmentation.How to provide personalized learning resources recommendation service for learners has become a challenge for online learning platform.There are a lot of historical learning data in online learning platform,and a lot of potential information can be obtained in online learning resources.Therefore,it is feasible to construct personalized learning resource recommendation method based on in-depth learning.In view of the above problems,the following work has been done in this study:(1)The research status of personalized Resource Recommendation Technology in online learning platform at home and abroad is summarized and analyzed.Then,from the perspective of Personalized Learning Resource Recommendation research,a comprehensive literature review is made and related technologies are summarized,which provides theoretical guidance and technical support for the development of this research work.(2)A multi-mode collaborative dominated personalized learning resource recommendation method is proposed.Firstly,a deep neural network input optimization strategy based on mutual information feature selection model MIFS is designed,and the output visual description under the learner-resource bipartite graph association model is established.Secondly,the deep neural network training is used to obtain the Resource Recommendation Model to realize personalized learning resource recommendation.The experimental results show that the recommendation method has good adaptability and performance.(3)On the basis of the above-mentioned methods,considering that learners' interests may change with time,a personalized learning resource recommendation method based on feature weighting optimization is proposed.This method focuses on several important learner-learning resource association features and regards recommendation as a multi-objective problem.For this purpose,several sub-objective functions are constructed and adopted.The weights of the dynamic adjustment features of the deep neural network are assigned to the sub-objective function,and the linear weights are the final multi-objective function.Finally,the function is solved and the recommendation sequence of personalized learning resources is obtained.The experimental results show that the method is feasible,efficient and stable.(4)In order to show the effect of the recommendation method more intuitively,the recommendation platform was initially implemented,and a deep neural network algorithm is realized.The architecture design,data association and the result display of the recommendation module of the recommendation platform were elaborated.It was found that the recommendation results were more suitable for learners to carry out online learning.Using deep learning algorithm to analyze and study learner preferences and learner's characteristic types,it can predict learner preferences accurately and improve learner's online learning experience and effect.Therefore,it is of practical significance and necessity to study how to apply personalized recommendation to online learning platform.