Study on Academic Documents _Oriented Automatic Summarization of Short Texts

2nd International Symposium on Application of Materials Science and Energy Materials

材料科学;能源学

Gao, Chunxiao^1 ; Guan, Bo^1 ; Zhang, Xiaoyue^1 ; Liu, Hao^1^2 ; Wei, Zhiqiang^1^2

Department of Computer Science and Technology, Ocean University of China, Qingdao, China^1

Pilot National Laboratory for Marine Science and Technology, Qingdao, China^2

关键词: Academic literature;    Attention mechanisms;    Automatic summarization;    Automatic text summarization;    Chinese literature;    Document summarization;    Reliable documents;    Text summarization;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/490/6/062025/pdf
DOI  :  10.1088/1757-899X/490/6/062025

学科分类：材料科学（综合）

来源: IOP

【 摘 要 】

Traditional automatic text summarization relies heavily on the original text information, and the extensibility is limited. However, generation-style abstractive methods attempt to generate the corresponding summarization by understanding the original semantics. We set out to set up a sequence-to-sequence model for academic document summarization generation. For purpose of reducing the detail loss of input sequence information, we put forward the attention mechanism to assign the weight of each input word. We trained this model on Chinese literature data set. It generated a reliable document summary. Our test shows that the approach has good adaptability to Chinese academic literature and has good performance in text summarization.