Improving the Robustness of Summarization Systems with Dual Augmentation

TL;DR

This paper investigates the robustness of summarization models against input perturbations and noise, proposing dual data augmentation techniques to enhance their resilience and improve performance on noisy and adversarial datasets.

Contribution

It introduces SummAttacker for generating semantic-consistent adversarial samples and a manifold softmixing method to augment training data, significantly boosting robustness of summarization models.

Findings

Models' performance drops significantly on adversarial inputs.

Data augmentation improves robustness on noisy and attacked datasets.

Proposed methods outperform strong baselines on Gigaword and CNN/DM datasets.

Abstract

A robust summarization system should be able to capture the gist of the document, regardless of the specific word choices or noise in the input. In this work, we first explore the summarization models' robustness against perturbations including word-level synonym substitution and noise. To create semantic-consistent substitutes, we propose a SummAttacker, which is an efficient approach to generating adversarial samples based on language models. Experimental results show that state-of-the-art summarization models have a significant decrease in performance on adversarial and noisy test sets. Next, we analyze the vulnerability of the summarization systems and explore improving the robustness by data augmentation. Specifically, the first brittleness factor we found is the poor understanding of infrequent words in the input. Correspondingly, we feed the encoder with more diverse cases created by SummAttacker in the input space. The other factor is in the latent space, where the attacked inputs bring more variations to the hidden states. Hence, we construct adversarial decoder input and devise manifold softmixing operation in hidden space to introduce more diversity. Experimental results on Gigaword and CNN/DM datasets demonstrate that our approach achieves significant improvements over strong baselines and exhibits higher robustness on noisy, attacked, and clean datasets.