Improvements in Multi-Document Abstractive Summarization using Multi Sentence Compression with Word Graph and Node Alignment

作者：

Highlights：

• Word Graph based representation of clusters of similar sentences.

• The nodes of the Word Graph are aligned to fuse multiple chunks of information.

• Similar sentences are compressed by traversing between fixed nodes of the graph.

• Integer Linear Programming based maximization of grammaticality and informativeness.

• Improved results are obtained for Sentence Fusion and Multi-Document Summarization.

摘要

•Word Graph based representation of clusters of similar sentences.•The nodes of the Word Graph are aligned to fuse multiple chunks of information.•Similar sentences are compressed by traversing between fixed nodes of the graph.•Integer Linear Programming based maximization of grammaticality and informativeness.•Improved results are obtained for Sentence Fusion and Multi-Document Summarization.

论文关键词：Multi-document abstraction,Word Graph,Node alignment,Multi sentence compression,Sentence fusion

论文评审过程：Received 21 April 2020, Revised 10 September 2021, Accepted 23 October 2021, Available online 13 November 2021, Version of Record 17 November 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116154