Part-of-Speech Tagging of Parallel Corpus based on Convolutional Neural Network

Parallel corpora part-of-speech tagging is an indispensable part in the application field of natural language information processing. The core work of this paper is to build a mathematical model of part-of-speech tagging based on CNNbased parallel corpus, which includes three aspects. Firstly, the basic CNN model is studied. Based on the artificial neuron model, the convolutional layer, the pooled layer and the fully connected layer are studied. Then, we study the maximum entropy model, integrate various information into one model, and carry out parts-of-speech tagging and syntactic analysis to improve the accuracy of natural language processing. Finally, the learning process of CNN is studied, the output is obtained by using forward propagation, and the error is fed back to the trainable parameters of the network through back propagation to realize parameter update. On this basis, NIST corpus is selected to compare the model proposed in this paper with the model commonly used in CNN, which proves that the model proposed in this paper has advantages in the accuracy of part-of-speech tagging.