Yong Guan, Shaoru Guo, Ru Li, Xiaoli Li and Hu Zhang, Integrating Semantic Scenario and Word Relations for Abstractive Sentence Summarization, EMNLP 2021
Recently graph-based methods have been adopted for Abstractive Text Summarization. However, existing graph-based methods only consider either word relations or structure in005 formation, which neglect the correlation be006 tween them. To simultaneously capture the word relations and structure information from sentences, we propose a novel Dual Graph net009 work for Abstractive Sentence Summarization (DG-ABS). Specifically, we first construct se011 mantic scenario graph and semantic word relation graph based on FrameNet, and subse quently learn their representations and design
graph fusion method to enhance their correla015 tion and obtain better semantic representation for summary generation. Experimental result017 s show our model outperforms existing state018 of-the-art methods on two popular benchmark datasets, i.e., Gigaword and DUC 2004.
This research is supported by core funding from: Institute for Infocomm Research
Grant Reference no. : Nil