Research On Automatic Text Classification Of Educational Technology Academic Paper Based On Deep Learnig

The development of educational technology is accompanied by an ever-increasing number of educational technology researchers and academic papers and researchers are increasingly looking for related literature to read.learn,and reference on-line.However,current literature database platforms only classify document types.Including review literature and policy research category 2,which causes researchers in the process of reading the literature is very inconvenient.The deep learning is developing,more and more classify texts are usually use neural network models,and deep learn applicating field is expanding.It becomes possible to classify texts of educational technology academic papers.Text categorization allow researchers obtain the direction of a discipline's information or disciplines better.Therefore,for the purpose of solve inconvenient retrieval of educational technology researchers in massive literature databases,this paper advance assorting of the study on educational technology academic papers based on deep learning methods,and explores the methods applicable to the text classification of educational technology academic papers and adopts them.This method classifies the 2008-2017 educational technology journal papers and analyzes the classification results.The main contents of this study are:Combining the hot topics of educational technology papers and the classification of periodical's papers in the industry,the educational technology academic literatures arc divided into four research categories:education theory or education methods research,distance education and online education research,and learning resources and technologies,subject progress and education administration,and detailed descriptions and keyword of each category.When pre-processing texts of educational technology academic papers,an educational technology corpus was constructed and used for word segmentation of texts of educational technology academic papers.Through experimental research on the effects of training educational technology word vectors on the text classification effect of educational technology academic papers,and the result of differ neural network patterns on the text classification result of educational technology academic papers,the classification methods applicable to the text classification of educational technology academic papers are sought,and by the text outcome inferred conclusions:(1)The accuracy of text classification using Word2vec word vectors trained in vocabulary of educational technology is higher than the use of Word2vec word vectors that are randomly initialized.(2)Bi-GRU model is more accurate than Text-CNN model and Bi-LSTM model for text classification training.The accuracy rate is 79.3 5%.Therefore,it seems that after training word vectors in education domain,Bi-GRU is used.The text classification training should use the model that is a text classification method suitable for education technology's text category.Finally,this paper uses this method to classify the 2008-2017 educational technology journal papers and analyze the classification results.The conclusions are as follows:In the exploring educational technology journal literatures process,the research on educational theory and teaching methods and methods has been on a rising trend since 2014,and forecasting will also be a research hotspot in the next few years;for distance education and online education research,Overall,the research shows a downward trend,but there are fluctuations.forecasting that small fluctuations will occur in the next few years,showing a general decline tendency;the research of learning resources and technologies,discipline development,and teacher and student administration will be relatively stable,and forecasting that it will develop steadily in the next few years.