Abstract

A good document summary should summarize the core content of the text. Research on automatic text summarization attempts to solve this problem. The encoder-decoder model is widely used in text summarization research. Soft attention is used to obtain the required contextual semantic information during decoding. However, due to the lack of access to the key features, the generated summary deviates from the core content. In this paper, we proposed an encoder-decoder model based on a double attention pointer network (DAPT). In DAPT, the self-attention mechanism collects key information from the encoder, the soft attention and the pointer network generate more coherent core content, and the fusion of both generates accurate and coherent summaries. In addition, the improved coverage mechanism is used to address the repetition problem and improve the quality of the generated summaries. Simultaneously, scheduled sampling and reinforcement learning (RL) are combined to generate new training methods to optimize the model. Experiments on the CNN/Daily Mail dataset and the LCSTS dataset show that our model performs as well as many state-of-the-art models. The experimental analysis shows that our model achieves higher summarization performance and reduces the occurrence of repetition.

Highlights

  • Automatic text summarization is a technology that has evolved to conform to the development of the information age

  • We present a dual-attention pointer network (DAPT) model, using the self-attention mechanism to obtain the key features of the text from the encoder

  • In the study of existing models, we found that the pointer network [4] can reproduce the details of the facts well, and generate more coherent and accurate summaries through the context vector and the attention pointer

Read more

Summary

INTRODUCTION

Automatic text summarization is a technology that has evolved to conform to the development of the information age. On the basis of this work and the machine translation method proposed by Bahdanau et al [12], Nallapati et al [13] provided a seq2seq+attention baseline model and constructed a CNN/Daily Mail text summarization dataset. See et al [4] improved on the coverage model of Tu et al [24] to address the repetitive problem in text summarization by setting a coverage vector whose value is the sum of the attention distributions computed by all previous prediction steps. D. REINFORCEMENT LEARNING In the text summarization, the ‘‘exposure bias’’ and loss-evaluation mismatch problems common in the model can be solved by introducing the ROUGE indicator during training. Paulus et al [5] introduced SCST to text abstracts and integrated the ‘‘teacher forcing’’ algorithm to improve the quality of the generated abstract while solving the ‘‘exposure bias’’ and loss-evaluation mismatch problems. The improved coverage mechanism will reduce repetition by participating in soft attention calculations

SELF-ATTENTION
IMPROVED COVERAGE MECHANISM
CONCLUSION

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.