Semantic-Preserving Abstractive Text Summarization with Siamese Generative Adversarial Net
Xin Sheng, Linli Xu, Yinlong Xu, Deqiang Jiang, Bo Ren
Abstract
We propose a novel siamese generative adversarial net for abstractive text summarization (SSPGAN), which can preserve the main semantics of the source text. Different from previous generative adversarial net based methods, SSPGAN is equipped with a siamese semantic-preserving discriminator, which can not only be trained to discriminate the machine-generated summaries from the human-summarized ones, but also ensure the semantic consistency between the source text and target summary. As a consequence of the min-max game between the generator and the siamese semantic-preserving discriminator, the generator can generate a summary that conveys the key content of the source text more accurately. Extensive experiments on several text summarization benchmarks in different languages demonstrate that the proposed model can achieve significant improvements over the state-of-the-art methods.
BibTeX
@inproceedings{sheng-etal-2022-semantic,
title = "Semantic-Preserving Abstractive Text Summarization with {S}iamese Generative Adversarial Net",
author = "Sheng, Xin and
Xu, Linli and
Xu, Yinlong and
Jiang, Deqiang and
Ren, Bo",
editor = "Carpuat, Marine and
de Marneffe, Marie-Catherine and
Meza Ruiz, Ivan Vladimir",
booktitle = "Findings of the Association for Computational Linguistics: NAACL 2022",
month = jul,
year = "2022",
address = "Seattle, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-naacl.163/",
doi = "10.18653/v1/2022.findings-naacl.163",
pages = "2121--2132"
}