Enriching and Controlling Global Semantics for Text Summarization

• DOI: 10.18653/v1/2021.emnlp-main.744
• Publication Date: 2021-12-17T03:56:42Z
• Authors: Thong Nguyen; Anh Tuan Luu; Truc Lu; Tho Quan
• Keywords: Multi-document summarization; Representation

Recently, Transformer-based models have been proven effective in the abstractive summarization task by creating fluent and informative summaries. Nevertheless, these still suffer from short-range dependency problem, causing them to produce summaries that miss key points of document. In this paper, we attempt address issue introducing a neural topic model empowered with normalizing flow capture global semantics document, which are then integrated into model. addition, avoid overwhelming effect on contextualized representation, introduce mechanism control amount supplied text generation module. Our method outperforms state-of-the-art five common datasets, namely CNN/DailyMail, XSum, Reddit TIFU, arXiv, PubMed.