Research On Screening Method Of Dynamic Mathematics Resources Based On Machine Learning

With the continuous development of education informatization and the continuous advancement of education modernization,the construction and application of digital education resources are becoming more and more important.With the increasing development of the Internet,big data and artificial intelligence,various digital education resource platforms have emerged one after another,and the types of digital education resources have become diverse,Among the many digital education resources,dynamic mathematics resources are widely used in mathematics education because of their subject,interactive and interesting impact on mathematics education.With the continuous increase of the number of users,the number of resources increases sharply,and a large number of resources are misclassified,so in the massive dynamic mathematical resources,the correct classification of resources faces great challenges and is also an urgent problem to be solved.The screening of dynamic mathematical resources can be summarized as a binary classification problem,and this paper proposes a dynamic mathematical resources binary classification screening method based on machine learning,and applies this method to the resources screening of the mathematics education cloud platform "Network Drawing Board",which achieves good practical results.Specifically,the main work of this article is as follows:(1)A Dynamic Mathematics Resources Feature Extraction(DMRFE)method is proposed.Firstly,the basic components of dynamic mathematical resources are divided into two types: interactive and constrained,and the geometric features of the constraint elements are constructed by quantitative statistics.Secondly,the automation technology is used to simulate the interaction between the user and the resources,generate the resources operation video,apply the moving object detection technology to "clean" the video frame picture,obtain the outline of the main figure,and complete the recognition of the main figure through the geometric pattern recognition technology,so as to construct the image features.Finally,with the help of Chinese word segmentation,stop word filtering,TF-IDF and other technologies,the text in resource is vectorized,and the LDA model is used to mine the hidden information of the text and construct the text features.(2)Form a data preprocessing method for dynamic mathematical resources.Firstly,the missing values and outliers of the sample are processed to get an overview of the full picture of the data.Secondly,the data normalization process is carried out for the problem of inconsistent dimensional between features.Finally,the IV value and Pearson coefficient are used for feature selection.(3)Build a dynamic mathematical resources screening fusion model.Through the five machine learning classification algorithms of K-nearest neighbor algorithm,support vector machine SVM,decision tree DT,random forest RF and GBDT,the hyperparameters are determined by the grid search method to construct the basic model of resources screening.Using the Stacking method,different basic models are used to form a fusion model of resources screening,and based on the model,the resources screening of Net Pad platform is carried out,and the rationality and efficiency of the model screening are proved by two indicators: accuracy and time saving percentage.