Research On Personalized Recommendation Methods For Education Field

The rapid development of information technology and the concept of "Internet +" education innovation have made online education widely concerned.As the market size and learners scale of online education continue to grow,the learning resources on the Internet have increased dramatically,many learners have encountered some problems in online education,such as resource overload and learning disorientation.How to mine and recommend learning resources which are interesting to learners from the vast amount of learning resources,so as to achieve personalized learning has become the focus of research.The emergence of information retrieval has solved the above problems.While,this method requires the learner to passively acquire the learning resources,personalized learning resource recommendation system came into being at the historic moment.However,this method requires the learner to passively acquire the learning resources,thereby recommending the application of the system.At present,the individualized learning resources recommendation is in a initial stage,therefore,and there are still many problems.In this paper,the problems of Learning Resource Recommendation in the field of education are analyzed and improved through new recommendation technology,the main work of this paper are as follows:1.The traditional collaborative filtering recommendation method always neglects the minority resources,and therefore,its recommendation results are limited.In view of this,the physical theory including heat spreading and mass diffusion are introduced into the recommendation system,and based on which,the model of hybrid learning resources recommendation based on bipartite network is constructed.In this model,the recommendation method based on heat spreading and the recommendation method based on mass diffusion are both taken into consideration,and they play different parts in different cases by an adjustable parameter.Compared with the traditional collaborative filtering recommendation method show that material diffusion recommendation method and heat conduction recommendation method,the experimental results show that the improved hybrid recommendation method performs has better recommendation effect.2.The traditional learning resource recommendation algorithms do not analyze learner'sinterest preferences from multiple dimensions of learner interest,some problems such as insufficient description of the learner's interest and inefficient recommendation should be solved in the process of recommendation.In view of this,multi-dimensional interest model is introduced into recommendation system.In this model,that consists of two kinds of dimensions: interest characteristics and stability of interest,then formed four basic dimensions,namely short-term feature interest,short-term link interest,long-term feature interest and long-term link interest,according to which an interest model was built,and based on which,recommendation system is established.Experiment on datasets verify that learning resource recommendation method based on multi-dimensional interest has better recommendation effect.This paper is well defined and it provides a good example for personalized recommendation.