Research on Single Document Automatic Summarization Method Based on Hybrid Neural Network
Download as PDF
DOI: 10.23977/csic.2018.0959
Author(s)
Qiaohong Chen, Wen Dong, Qi Sun, Yubo Jia
Corresponding Author
Qiaohong Chen
ABSTRACT
In order to extract the required information from the massive information, an automatic summarization method based on hybrid neural network is proposed to help people to browse and understand the document quickly, and improve the efficiency of automatic summarization. The method combines the Convolutional Neural Network that has high efficiency and small over-fitting phenomenon in the training process and the Long Short-Term Memory Neural Network model with good effect on sequence prediction. The model takes full account of the characteristics words, characteristic sentences, feature segment positions and other factors, and add a signal to the input of the Long Short-Term Memory. Experimental results show that compared with automatic summarization methods based on LSI model, LDA model, TextRank summarization algorithm, PCA summarization algorithm and Long Short-Term Memory Neural Network model, the proposed method based on hybrid neural network has a good effect on automatic summarization, and improves the quality of automatic summarization effectively.
KEYWORDS
Hybrid neural network, Automatic summarization, Convolutional Neural Network, Long Short-Term Memory Neural Network, Deep learning