Re-evaluating Automatic Summarization with BLEU and 192 Shades of ROUGE
Open accessThe DOI matches its Crossref record; 3 of the 3 references it cites were found in a registry or an index.
Abstract
We provide an analysis of current evaluation methodologies applied to summarization metrics and identify the following areas of concern: (1) movement away from evaluation by correlation with human assessment; (2) omission of important components of human assessment from evaluations, in addition to large numbers of metric variants; (3) absence of methods of significance testing improvements over a baseline.We outline an evaluation methodology that overcomes all such challenges, providing the first method of significance testing suitable for evaluation of summarization metrics.Our evaluation reveals for the first time which metric variants significantly outperform others, optimal metric variants distinct from current recommended best variants, as well as machine translation metric BLEU to have performance on-par with ROUGE for the purpose of evaluation of summarization systems.We subsequently replicate a recent large-scale evaluation that relied on, what we now know to be, suboptimal ROUGE variants revealing distinct conclusions about the relative performance of state-of-the-art summarization systems.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BLEU: a method for automatic evaluation of machine translation
Papineni, K., Roukos, S., Ward, T., & Zhu, W.-J. (2001) · Proceedings of the 40th Annual Meeting on Association for Computational Linguistics - ACL '02 · doi:10.3115/1073083.1073135 · Cited by 21,344
Moses: open source toolkit for statistical machine translation
Koehn, P., Zens, R., Dyer, C., Bojar, O., Constantin, A., Herbst, E., Hoang, H., Birch, A., Callison-Burch, C., Federico, M., et al. (2007) · Proceedings of the 45th Annual Meeting of the ACL on Interactive Poster and Demonstration Sessions - ACL '07 · doi:10.3115/1557769.1557821 · Cited by 4,735
Tests for comparing elements of a correlation matrix
Steiger, J. H. (1980) · Psychological Bulletin · doi:10.1037/0033-2909.87.2.245 · Cited by 4,621 · cited 2×
“Williams test 3 (Williams, 1959) evaluates the significance of a difference in dependent correlations (Steiger, 1980).”
Background
Cited by
Recent automatic text summarization techniques: a survey
Gambhir, M., & Gupta, V. (2017) · Artificial Intelligence Review · doi:10.1007/s10462-016-9475-9 · Cited by 727
SummEval: Re-evaluating Summarization Evaluation
Fabbri, A. R., Kryściński, W., McCann, B., Xiong, C., Socher, R., & Radev, D. (2021) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00373 · Cited by 587
“In a similar line of work, Graham (2015) conducted a large-scale study of the effectiveness of different ROUGE metric variants and compared the it against the BLEU metric on the DUC datasets.”
ResultA Structured Review of the Validity of BLEU
Reiter, E. (2018) · Computational Linguistics · doi:10.1162/coli_a_00322 · Cited by 386
“Five of these correlations are Low; the sixth (reported in Graham (2015)) is High.”
BackgroundText Preprocessing for Text Mining in Organizational Research: Review and Recommendations
Hickman, L., Thapa, S., Tay, L., Cao, M., & Srinivasan, P. (2022) · Organizational Research Methods · doi:10.1177/1094428120971683 · Cited by 362
Neural Text Summarization: A Critical Evaluation
Kryscinski, W., Keskar, N. S., McCann, B., Xiong, C., & Socher, R. (2019) · Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) · doi:10.18653/v1/d19-1051 · Cited by 335
Evaluation of Text Generation: A Survey
Celikyilmaz, A., Clark, E., & Gao, J. (2020) · arXiv · doi:10.48550/arxiv.2006.14799 · Cited by 207
Understanding Factuality in Abstractive Summarization with FRANK: A Benchmark for Factuality Metrics
Pagnoni, A., Balachandran, V., & Tsvetkov, Y. (2021) · Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies · doi:10.18653/v1/2021.naacl-main.383 · Cited by 197
Re-evaluating Evaluation in Text Summarization
Bhandari, M., Gour, P. N., Ashfaq, A., Liu, P., & Neubig, G. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2020.emnlp-main.751 · Cited by 148
Repairing the Cracked Foundation: A Survey of Obstacles in Evaluation Practices for Generated Text
Gehrmann, S., Clark, E., & Sellam, T. (2023) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.13715 · Cited by 123
The limits of automatic summarisation according to ROUGE
Schluter, N. (2017) · Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers · doi:10.18653/v1/e17-2007 · Cited by 108
“Previous research on evaluation metrics for automatic summarisation has tried to empirically show a correlation between human judgments and system output quality (Lin, 2004; Lin and Hovy, 2003; Liu and Liu, 2008; Graham, 2015).”
BackgroundGO FIGURE: A Meta Evaluation of Factuality in Summarization
Gabriel, S., Celikyilmaz, A., Jha, R., Choi, Y., & Gao, J. (2021) · Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 · doi:10.18653/v1/2021.findings-acl.42 · Cited by 75
“Prior work concerning evaluation of automatic metrics and human evaluation for NLG systems has mainly focused on general analysis of output quality or coherence and fluency (Callison-Burch et al., 2007; Graham, 2015; Fabbri et al., 2021), rather than factuality.”
BackgroundA Dataset and Evaluation Metrics for Abstractive Compression of Sentences and Short Paragraphs
Toutanova, K., Brockett, C., Tran, K. M., & Amershi, S. (2016) · Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d16-1033 · Cited by 42
CrystalBLEU: Precisely and Efficiently Measuring the Similarity of Code
Eghbali, A., & Pradel, M. (2022) · Proceedings of the 37th IEEE/ACM International Conference on Automated Software Engineering · doi:10.1145/3551349.3556903 · Cited by 42
“Graham [14] presents a detailed analysis of different variants of ROUGE and BLEU with regards to their correlation with the human judgement.”
BackgroundOn the Limitations of Reference-Free Evaluations of Generated Text
Deutsch, D., Dror, R., & Roth, D. (2022) · arXiv · doi:10.48550/arxiv.2210.12563 · Cited by 29
QAScore—An Unsupervised Unreferenced Metric for the Question Generation Evaluation
Ji, T., Lyu, C., Jones, G., Zhou, L., & Graham, Y. (2022) · Entropy · doi:10.3390/e24111514 · Cited by 12
Self-citation
Metadata
- Authors
- Graham, Y.
- Venue
- Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing
- Published
- 2015
- Pages
- 128-137
- OpenAlex
- W2251023345
- Cited by
- 143
- References
- 3
- FWCI
- 4.36
- Type
- conference paper
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.