Text Summarization Model of Combining Global Gated Unit and Copy Mechanism

Abstract

Text summarization is a common task in NLP. Automatic text summarization aims to transform lengthy documents into shortened versions. Recently, the neural networks based on seq2seq with attention are good at generating summarization. However, the accuracy of the summarization too difficult are to guarantee. In addition, the Out-of-Vocabulary (OOV) problem is also an important factor affecting the quality of the generated summary. To solve these problems, we hybrid the advantages of the extractive and abstractive summarization systems to propose text summarization model of combining global gated unit and copy mechanism (GGUC). The experiment results demonstrate that the performance of the model is better than the other text summary system on LCSTS datasets.