基于伪反馈蒸馏的无监督双级联查询式抽取摘要学习
计算与语言
2018-11-02 v1
摘要
我们提出 Dual-CES——一种新颖的无监督、查询聚焦、多文档抽取式摘要器。Dual-CES 旨在更好地处理摘要中显著性与聚焦之间的权衡。为此，Dual-CES 采用一种基于显著性的伪反馈蒸馏的两步双级联优化方法。总体而言，Dual-CES 显著优于所有其他最先进的无监督替代方法。Dual-CES 甚至被证明能够优于强监督摘要器。
引用
@article{arxiv.1811.00436,
title = {Unsupervised Dual-Cascade Learning with Pseudo-Feedback Distillation for Query-based Extractive Summarization},
author = {Haggai Roitman and Guy Feigenblat and David Konopnicki and Doron Cohen and Odellia Boni},
journal= {arXiv preprint arXiv:1811.00436},
year = {2018}
}