Research On Learning Resource Recommendation Technology In Online Education

With the deep integration of digital information technology and intelligent education,remote network education has become an indispensable part of modern education.The emergence of online education platforms has provided new ideas and new ways to meet people’s growing demand for educational resources and facilitated people’s access to educational resources.However,in the era of big data,with the continuous expansion of the scale of educational resources,learners need to spend more time and energy to find the suitable learning resources,resulting in low resource utilization and learning efficiency.Therefore,learning resource recommendation technology based on online education came into being.However,the existing learning resource recommendation methods often have some limitations to a certain degree.Firstly,learners are the main body of online education.The lack of mining and analysis of learner-related features makes the algorithm unable to capture the actual needs of learners.Secondly,traditional recommendation scheme tend to make learners fall into the "information cocoon",which means that they will always learn similar or unchallenged content.Obviously,it is not beneficial to the development of learners’ intelligence.Thirdly,the recommended learning resources are often presented in the form of an unordered set,which fails to dynamically plan the concrete learning path for learners.Based on the above limitations,the learning resource recommendation technology in the field of online education is studied in this paper.A learner model is constructed to extract personalized features of learners,and it serves as the design basis for the educational video resource recommendation algorithm.Furthermore,the learning path recommendation algorithm is developed to provide an orderly and logical learning path for learners to maximize their knowledge level.The research work of this paper is mainly reflected as follows:First,this paper constructs a learner model in the scenario of online learning.First of all,the learner model is analyzed hierarchically,and the composition and functions of data layer,analysis layer,presentation layer and application layer are described in detail.Then,the field information and processing process of the dataset are introduced,and various learning behaviors are screened and analyzed.Finally,feature extraction is carried out for learners.Specifically,the ability characteristic of learners is calculated based on the theory of educational psychology,and the machine learning algorithm is used to classify the learner identity according to their ability and participation.The learning pattern and rule of each learner group is summarized according to above features.Second,this paper proposes the educational video resource recommendation algorithm based on personalized exploration strategy.Firstly,the features of educational video learning resources are extracted based on learners’ behavior records to enhance their flexibility.Secondly,the method improves the contextual Lin UCB recommendation algorithm from two aspects of parallel matrix calculation and personalized exploration strategy,which realizes the improvement of algorithm efficiency and adaptive adjustment for the ratio of exploration and utilization.Through this improvement,the difficulty of educational videos can be controlled within the range of learners’ abilities,and the potential of learners can be exploited at the same time,so as to reduce the risks brought by the exploration process.Finally,the experimental results show that the performance of this method is improved in terms of accuracy,personalization and adaptability.Last but not least,this paper develops the learning path recommendation algorithm based on attentive knowledge tracing.First of all,a knowledge tracing model based on attention mechanism is constructed.The improved embedding representation method is used to enhance the interpretability of the model,which is designed based on the dual parameter logistic model.At the same time,the forgetting behaviors of learners are considered to track learners’ knowledge level more accurately.Secondly,a search space optimization algorithm based on knowledge map is proposed to reduce the execution scope of the algorithm and ensure the logic of the learning path,and then prevent deviation from learning objectives.Then,a learning path recommendation algorithm is designed based on a variety of constraint rules.Specifically,the learning content is gradually recommended according to the rules of interpretability,rationality and effectiveness in the concept change model,and a dynamic learning path is formed over time eventually.Finally,the experimental results show that the attentive knowledge tracing algorithm has significantly improved the prediction accuracy compared with the traditional knowledge tracing method,and the proposed learning path recommendation algorithm can also optimize the overall effect of the learning path.