Algorithm of Part-of-Speech Tagging of Corpus Based on Recurrent Neural Network

Deep learning uses multi-layer neural networks to learn and understand complex data algorithms, which greatly taps the potential value of massive data. Part-of-speech tagging is one of the basic tasks in the field of natural language processing, and the tagging effect has a great impact on subsequent research tasks such as syntactic analysis, information extraction and machine translation. In the field of corpus part-of-speech tagging, RNN has obvious advantages over traditional statistical machine learning methods in terms of tagging accuracy and speed. Based on the theory and method of deep learning and RNN, this paper firstly studies the part-of-speech tagging language model and algorithm, including N-gram language model, language model evaluation method and parameter smoothing algorithm. Then, the annotation process based on recurrent neural network is studied, and the backpropagation algorithm of RNN in corpus annotation is deeply studied. Finally, simulation experiments are carried out, and the results show that the part-of-speech tagging algorithm based on RNN has obvious advantages over other deep learning models.