QuestEval: Summarization Asks for Fact-based Evaluation - Sorbonne Université
Conference Papers Year : 2021

QuestEval: Summarization Asks for Fact-based Evaluation

Abstract

Summarization evaluation remains an open research problem: current metrics such as ROUGE are known to be limited and to correlate poorly with human judgments. To alleviate this issue, recent work has proposed evaluation metrics which rely on question answering models to assess whether a summary contains all the relevant information in its source document. Though promising, the proposed approaches have so far failed to correlate better than ROUGE with human judgments. In this paper, we extend previous approaches and propose a unified framework, named QuestEval. In contrast to established metrics such as ROUGE or BERTScore, QuestEval does not require any ground-truth reference. Nonetheless, QuestEval substantially improves the correlation with human judgments over four evaluation dimensions (consistency, coherence, fluency, and relevance), as shown in extensive experiments.
Fichier principal
Vignette du fichier
2021.emnlp-main.529.pdf (578.62 Ko) Télécharger le fichier
Origin Publisher files allowed on an open archive

Dates and versions

hal-03541895 , version 1 (25-01-2022)

Identifiers

Cite

Thomas Scialom, Paul-Alexis Dray, Patrick Gallinari, Sylvain Lamprier, Benjamin Piwowarski, et al.. QuestEval: Summarization Asks for Fact-based Evaluation. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, Nov 2021, Punta Cana (en ligne), Dominican Republic. pp.6594-6604, ⟨10.18653/v1/2021.emnlp-main.529⟩. ⟨hal-03541895⟩
85 View
111 Download

Altmetric

Share

More