Chinese Text Classification Based On Deep Learning

With the development of big data technology,the amount of data faced by text analysis is growing rapidly.Processing this data and obtaining important information from it for managers to analyze and make decisions are of great significance to social development.Therefore,NLP(Natural Language Processing)is receiving more and more attention.Currently,text classification technology is gradually developing from traditional methods based on statistics and machine learning to deep learning methods based on complex neural network structures.However,the traditional deep learning structures have many problems and potentials for text processing.This paper studies the theories of various deep learning models and their applications in NLP,and proposes the use of improved recurrent neural networks and convolutional neural networks in text classification.The contents and results of the research are as follows:1.This paper introduce NLP research at home and abroad by literature research.The basic process and concept of text classification are introduced in detail,including word segmentation method,word vector generation algorithm and common deep learning model theories.2.A new word vector generation method,the latest ELMo model,is studied and modified,and applied to the Chinese text classification task,and compared with the traditional word2 vec method.3.For the Chinese text classification model,this paper proposes a new deep learning model based on convolutional neural network and recurrent neural network.By combining GRU and GCNN models,the text information is extracted by recurrent layers and convolutional layers.The training time of the original LSTM structure was shortened by the GRU structure,and the classification accuracy rate on Sogou news data reached 95.83%,which proved the validity of the method compared with other comparison models.