A Text Classification Based On The Recurrent Neural Networks

With the advent of the big data era, all kinds of data in the society is growing at a rapid speed, so the text mining is becoming a problem that must be solved. Among all text mining techniques, the text classification is the foundation. Thus how to classify text automatic is worth to research. And based on the Recurrent Neural Networks which is one of the deep learning networks to mine the text data can solve the defect that ignoring the context information based on the statistical learning method. Also it would be a classic case in the future.This paper firstly introduces the general process of the text classification, and then analyzes every step during the text classification. As far as text representation model, we importantly introduce Vector Space Model. As far as the choice of classifier, we introduce Support Vector Machine. In terms of the defect of using the statistics learning to classify text data, we raise the neural network model to classify text data. This method can solve the defect that ignoring the context information. Thus it lead a new neural work-Recurrent Neural Networks, and then we infer the method of training Recurrent Neural Networks which called BPTT algorithm. Finally, the result of the experiment indicate the text classification using the Recurrent Neural Networks have higher recall number and precision number than statistical learning method.