Recognition of Parallelism Sentence Based on Recurrent Neural Network

Parallelism sentence is a group of sentence pairs with similar structure, same tone and similar semantics. Generally speaking, it consists of two or more sentences. Parallelism sentence have distinctive features such as compact structure, neat sentence patterns, expressiveness, etc. Parallelism sentence enhance language momentum and better express people thoughts and feelings. Widely used in various styles. The usage of the parallelism sentence can be used as a criterion for evaluating the article. Therefore, the identification of the paraphrase sentence is an important work in the text structure analysis. The current method is mainly based on feature extraction. Identifying from the grammar rules, structure and the degree of overlap. This type of method depends on the characteristics of artificial selection and does not take into account the content level of the sentence. Lack of understanding of the meaning of sentences. Therefore, the parallelism sentence recognition method based on recurrent neural network is proposed in this paper. The method does not require artificial design features and can avoid adverse effects on improper selection of features. At the same time, it can also consider the semantic meaning of sentences. The test results show that compared with similar researches, the accuracy of this method has greatly improved.