Classification of Discourse Role Based on Neural Network
Zhimin Chen, Wei Song, Lizhen Liu, Xinlei Shelly Zhao, Chao Du · 2017
In the era of the rapid development of today's culture, education has been paid more and more attention by society, and it is becoming more and more important to improve students' writing ability. In order to improve students' writing ability, students should learn the structure of essay, which is a key step in improving writing ability. Each sentence in essay has a corresponding role. We will identify the role of sentence and classify it. In this paper, we use the Tensorflow as the backend of the keras, we construct a neural network model, which mainly applys convolution neural network to identify the roles of the sentences in the english essay. Firstly, we find out the roles of the sentences. And according to the existing roles, we annotate the sentences of training set. Then, we turn each sentence into a corresponding vector matrix. Finally, the role of the sentence is identified and classified by our neural network model. The method proposed by us to identify the role of english sentence based on neural network can achieve the automatic identification and classification of sentences, and can help students learn the structure of an essay, and can help improve their writing ability.