Detecting and Mitigating Hallucinations in Multilingual Summarisation

TL;DR

This paper introduces mFACT, a new metric for evaluating faithfulness in multilingual summarisation, and proposes a loss weighting method to reduce hallucinations, significantly improving performance and faithfulness in cross-lingual transfer.

Contribution

It develops mFACT, the first faithfulness metric for non-English summaries, and presents a simple loss weighting approach to mitigate hallucinations in multilingual summarisation.

Findings

mFACT effectively detects hallucinations in multilingual summaries.

Loss weighting improves summarisation faithfulness and performance.

Method outperforms strong baselines like MAD-X in experiments.

Abstract

Hallucinations pose a significant challenge to the reliability of neural models for abstractive summarisation. While automatically generated summaries may be fluent, they often lack faithfulness to the original document. This issue becomes even more pronounced in low-resource settings, such as cross-lingual transfer. With the existing faithful metrics focusing on English, even measuring the extent of this phenomenon in cross-lingual settings is hard. To address this, we first develop a novel metric, mFACT, evaluating the faithfulness of non-English summaries, leveraging translation-based transfer from multiple English faithfulness metrics. We then propose a simple but effective method to reduce hallucinations with a cross-lingual transfer, which weighs the loss of each training example by its faithfulness score. Through extensive experiments in multiple languages, we demonstrate that mFACT is the metric that is most suited to detect hallucinations. Moreover, we find that our proposed loss weighting method drastically increases both performance and faithfulness according to both automatic and human evaluation when compared to strong baselines for cross-lingual transfer such as MAD-X. Our code and dataset are available at https://github.com/yfqiu-nlp/mfact-summ.