Improving the Faithfulness of Abstractive Summarization via Entity Coverage Control

TL;DR

This paper introduces Entity Coverage Control (ECC), a method to improve faithfulness in abstractive summarization by guiding models to better recognize and include entities, reducing hallucinations in both supervised and zero-shot settings.

Contribution

The paper proposes ECC, a novel entity-level control technique that enhances faithfulness in summarization models, including a zero-shot extension using intermediate fine-tuning on noisy Wikipedia data.

Findings

ECC improves faithfulness and salience in summarization.

The method outperforms baselines on XSum, Pubmed, and SAMSum datasets.

ECC reduces entity hallucinations in both supervised and zero-shot scenarios.

Abstract

Abstractive summarization systems leveraging pre-training language models have achieved superior results on benchmark datasets. However, such models have been shown to be more prone to hallucinate facts that are unfaithful to the input context. In this paper, we propose a method to remedy entity-level extrinsic hallucinations with Entity Coverage Control (ECC). We first compute entity coverage precision and prepend the corresponding control code for each training example, which implicitly guides the model to recognize faithfulness contents in the training phase. We further extend our method via intermediate fine-tuning on large but noisy data extracted from Wikipedia to unlock zero-shot summarization. We show that the proposed method leads to more faithful and salient abstractive summarization in supervised fine-tuning and zero-shot settings according to our experimental results on three benchmark datasets XSum, Pubmed, and SAMSum of very different domains and styles.