Research On News Text Classification Method Based On Deep Learning Model

In today’s society,the amount of information is increasing,and the news information is becoming more and more complex and diverse.For the task of news text classification,the correct classification is of great significance.Efficient and accurate classification of news can help users better understand the real situation and background of events,and also help users more quickly obtain the information they are interested in.In this paper,deep learning technology is used to classify news texts.Compared with traditional machine learning algorithms,deep learning has obvious advantages in text classification tasks.It can better deal with nonlinear relations in text data,and can automatically learn more complex feature representations.Therefore,this paper applies deep learning in the field of news text classification to improve the classification accuracy.The main work contents are as follows:(1)This paper reviews and analyzes the existing deep learning text classification methods,designs and implements new text classification algorithms.The Convolutional neural network can extract local features of text,but can’t capture structure information or semantic relationships between words,and a single CNN model’s classification performance is low,whereas GRU can effectively extract semantic information and global structure relationships of text.To address this problem,this paper proposes a news text classification method based on the GRU＿CNN model,which combines the advantages of CNN and GRU.The experimental results reveal that the GRU＿CNN hybrid model outperforms single CNN,LSTM,and GRU models in terms of classification effect and accuracy.(2)In order to solve the problem of low accuracy when Softmax is used for regression analysis in convolutional neural networks,a GRU＿CNN classification model with XGBoost classifier is proposed based on GRU＿CNN model.Experiments show that this model can further improve the classification accuracy and achieve better classification results,which provides certain application value in the field of news text classification.In the process of experiment,multiple groups of comparison experiments were conducted to verify the effectiveness of the model.In order to verify the effectiveness of the introduction of XGBoost classifier,several models were selected for comparative experiments,such as: CNN-SVM,CNN-NB and other models.In order to avoid overfitting,gradient vanishing and gradient explosion during model training,an effective Adam optimization algorithm is designed in the model and the dropout layer is introduced to reduce the overfitting phenomenon.The classification accuracy of the GRU＿CNN model designed in this paper is 97.86% on the Cnews dataset,and the classification accuracy of the further improved GRU＿CNN＿XGBoost model reaches 98.92%.