News Classification Of Deep Learning And SVM Fusion Technology Research

With the rapid development of information technology,people have stepped into a new stage.However,each person has limited time to focus on the news and can not quickly move from messy news.News classification is a technology that groups news in a way that automates the classification of newly generated news according to the area of news content.Users can quickly find the information they need according to the news categories.News classification belongs to a supervised machine learning technology.The algorithm needs to train the models according to the news content that has been labeled the correct category.In this paper,author design a news annotation system according to the requirements of news category annotation.In order to ensure the quality of annotation,a set of annotation specification and annotation validation methods are designed.This paper studies the traditional text classification algorithm,synthetically analyzes the principle and application range of the algorithm,and selects the SVM algorithm suitable for news classification.Based on SVM algorithm,an automatic news classification architecture based on SVM is designed,and the design idea of each module in the architecture is introduced in detail.Aiming at the characteristics of SVM classification algorithm,this paper improves the index of algorithm model,selects the key features of the features needed by algorithm calculation,and continuously iterates the feature extraction,selects the features that contribute the most to the classification,and improves the speed of the model.Then according to the status quo and trend that the proportion of the pictures in the news is larger and the proportion of the text is less and less,the deep learning method is integrated into the SVM classification,the deep features of the pictures in the news are extracted by using the convolutional neural network technique,the classification index has been improved to a certain extent.