Controlling Decoding for More Abstractive Summaries with Copy-Based Networks
Noah Weber 1, Leena Shekhar 1, Niranjan Balasubramanian1, Kyunghyun Cho2 1Stony Brook University, NY {nwweber, lshekhar, 2New York University, NY *These authors contributed equally to this work.
Abstract
Attention-based neural abstractive summarization systems equipped with copy mechanisms have shown promising results. Despite this success, it has been noticed that such a system generates a summary by mostly, if not entirely, copying over phrases, sentences, and sometimes multiple consecutive sentences from an input paragraph, effectively performing extractive summarization. In this paper, we verify this behavior using the latest neural abstractive summarization system - a pointer-generator network (See et al., 2017). We propose a simple baseline method that allows us to control the amount of copying without retraining. Experiments indicate that the method provides a strong baseline for abstractive systems looking to obtain high ROUGE scores while minimizing overlap with the source article, substantially reducing the n-gram overlap with the original article while keeping within 2 points of the original model’s ROUGE score.
中文速览
所谓"神经抽象摘要"系统在实践中其实大量直接复制原文句子，本质上做的是抽取式摘要，而非真正的抽象改写。研究者以 See 等人（2017）的指针生成网络（pointer-generator network）为对象，通过统计生成摘要与原文之间的 n-gram 重叠率，定量证实了这一现象——例如超过 20% 的 25-gram 片段与原文完全相同，而人工参考摘要几乎没有这类重叠。为此，研究者提出一种无需重新训练的简单改进方案：在束搜索（beam search）的打分函数中加入惩罚项，让解码过程中"生成"与"复制"的混合系数向预设目标靠拢，从而在解码阶段灵活控制抽象程度。实验结果显示，该方法将 25-gram 重叠率从约 29% 大幅压缩至约 4%，同时 ROUGE 分数仅下降不到 2 分，为后续研究"如何让摘要更真正抽象化"提供了一条清晰可复现的基线。
原文 arXiv:1803.07038;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1803.07038v2