基于联合注意力生成对抗网络的自动文摘模型  

Automatic summarization model based on joint attention generative adversarial network

作　　者：董张慧雅 张凡 王莉 DONG Zhang-hui-ya;ZHANG Fan;WANG Li(College of Information and Computer,Taiyuan University of Technology,Taiyuan 030600,China;College of Software,Taiyuan University of Technology,Taiyuan 030600,China;College of Data Science,Taiyuan University of Technology,Taiyuan 030600,China)

机构地区：[1]太原理工大学信息与计算机学院,山西太原030600 [2]太原理工大学软件学院,山西太原030600 [3]太原理工大学大数据学院,山西太原030600

出　　处：《计算机工程与设计》2021年第6期1756-1762,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61872260)。

摘　　要：针对现有自动文摘方法难以准确捕捉高信息量单词的问题,提出一种基于联合注意力的生成对抗网络模型并应用于文本的自动摘要。利用联合注意力机制构建新的单词层注意力,得到文档的上下文向量表示。提出一种联合损失函数训练生成器,最终生成摘要。实验结果表明,该模型与几种主流模型相比,ROUGE-1和ROUGE-L值均有所提高,表明该模型可有效提高自动文摘的质量。Since the existing automatic summarization method is difficult to accurately capture the words with high information,a computational model based on joint attention generative adversarial network was proposed,which was applied in automatic text summarization.Joint attention mechanism was used to build new word-level attention,and context vector was embedded.After the generator was updated according to the proposed joint loss function,the summary was generated.Compared with mainstream models,the values of ROUGE-1 and ROUGE-L of the proposed model are improved,indicating that it can effectively improve the quality of automatic summarization.

关 键 词：生成对抗网络 自动文摘 自然语言处理 分层注意力 编码器-解码器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]