Research And Application Of Text Classification Technology Based On Deep Learning

In the research field of natural language processing,text classification is a hot research direction.As a technology of information processing,more and more scholars lay emphasis on this focus.In recent years,due to the rapid rise of deep learning,and text classification methods based on deep learning gradually occupy the mainstream research direction.The news text is easy to obtain,large amount of data and complex content.The text classification technology applied to news text classification can effectively organize and manage these data.For the purpose of further improve the accuracy of classification,a new text classification algorithm based on word cooccurrence and graph convolution is proposed on text classification technology based on deep learning,and it is applied to news text classification.The main work and innovations in this paper are as follows:1.Summarize the text classification technology based on deep learning.To the problem of insufficient model generalization ability and low classification accuracy in the text classification algorithm based on graph convolutional neural network,a classification algorithm based on word co-occurrence and graph convolution is proposed.This algorithm integrates the attention mechanism into the model,which enhances the text representation.Experimental results show that compared with other traditional classification algorithms.The new algorithm can achieve better results in three different text classification tasks.2.A news text classification algorithm based on deep learning is proposed.The new algorithm applies the classification algorithm based on word co-occurrence and graph convolution to news text classification.Compared with other algorithm models,the new algorithm has good generalization ability,and can achieve better classification results,which provide certain application value for the field of news text classification.3.Explain the proposed model.Analyze the reasons for the better performance of the model on certain data sets.Through the design of multiple sets of comparative experiments,it is found that the label information of the document node can be passed to the adjacent word node in new model,which makes that the word node collects comprehensive document label information and spreads the label information to the entire text graph to complete the classification.It is proved the new model can effectively enhance the text representation.