Pith. sign in

REVIEW 4 major objections 6 minor 42 references

Can summarization approximate simplification? A gold standard comparison

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Paragraph-wise summarization approximates gold simplification (ROUGE-L 0.654, mean 0.566), while document-wide summarization stays below 0.141.

desk verdict Useful descriptive comparison, but the headline ROUGE-L lacks a source-text baseline, so the approximation claim is not yet supported. read the letter →

arxiv 2501.16181 v1 pith:QPGEWORW submitted 2025-01-27 cs.CL

Giacomo Magnifico , Eduard Barbu This is my paper · ORCID
classification cs.CL
keywords textsimplificationsummarizationBRIOROUGE-LNewselacorpusabstractiveparagraph-by-paragraphgoldstandardcomparison
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether a strong abstractive summarizer can stand in for manual text simplification. It runs the BRIO summarizer over 1,913 original Newsela articles in two modes, whole-document and paragraph-by-paragraph, and compares the outputs with the corpus's four professionally simplified levels using ROUGE-L. The paragraph mode reaches a top ROUGE-L of 0.654 against level-1 simplification and averages 0.566 across levels, while the document-wide mode stays between 0.109 and 0.141. The authors conclude that automated summarization does not replace simplification but produces text similar enough to serve as a starting baseline for simplification work. This result matters because simplification lacks standardized data, metrics, and training resources, and a lexical proxy could help bootstrap annotation and evaluation.

What carries the argument

The load-bearing mechanism is paragraph-by-paragraph segmentation feeding an abstractive summarizer. The summarizer is BRIO, a system that fine-tunes two generation models, one BART-based for long documents and one PEGASUS-based for short texts; this paper uses the BART variant. Splitting each article into paragraphs, summarizing each block separately, and rejoining the outputs preserves the topic structure that a whole-document pass condenses away. Similarity is measured with ROUGE-L, a longest-common-subsequence F1 score that rewards coherent word sequences and keyword retention, and this statistic is what produces the 0.654 result.

What would settle it

Measure ROUGE-L between the original Newsela articles and their simplified levels; if that score is comparable to 0.654, then source-text retention explains the similarity. Separately, rerun the paragraph-level outputs with a semantic metric such as ROUGE-SEM or BERTScore; if semantic similarity is low while lexical overlap is high, summarization is not approximating simplification, only its surface wording.

Watch

Extended reading notes

Core claim

The paper's central claim is that abstractive summarization, applied paragraph by paragraph, is a substantially better lexical approximation of gold-standard simplification than document-wide summarization is. On the Newsela corpus, the BART-based BRIO model reached ROUGE-L 0.654 at the mildest simplification level and a mean of 0.566 over all four levels, with scores decreasing as simplification becomes deeper; document-wide summarization scored only 0.109 to 0.141. The authors interpret the paragraph results as evidence that block-by-block processing mirrors how human annotators rewrite news text, and they conclude the output is similar enough to be used as a baseline for simplification, though not as a substitute for it.

Load-bearing premise

The argument depends on treating ROUGE-L lexical overlap with the gold simplifications as a measure of simplification-likeness, even though the score cannot detect semantically equivalent rewordings and no comparison is made against the original article or a length-matched baseline.

Editorial extensions

If this is right

  • Paragraph-by-paragraph summaries can provide a weak-label baseline for simplification projects: annotators start from a summary rather than from the full original article.
  • Document-wide summarization should not be used as a simplification proxy, because its ROUGE-L ceiling is below 0.141.
  • The usefulness of summarization as a simplification proxy decreases as the target reading level drops, since ROUGE-L falls monotonically from level 1 to level 4.
  • Re-evaluation with semantic metrics such as ROUGE-SEM or SARI, as the authors recommend, would test whether the lexical overlap reflects genuine simplification rather than surface wording.
  • The 10x to 50x runtime increase for paragraph-by-paragraph processing implies that practical deployment needs parallelization, lighter models, or optimization before large-scale use.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper's design does not compare the original Newsela article against the simplified levels, so part of the 0.654 overlap may come from words that both summarization and simplification inherit from the source; a source-overlap baseline would isolate the simplification-specific contribution.
  • A length-matched baseline is also missing: paragraph-wise summaries are longer than document-wise summaries, and longer outputs naturally share more tokens with the gold simplification, which could inflate ROUGE-L independently of simplification quality.
  • If semantic metrics confirm the convergence, paragraph-wise summarization could be used to generate cheap pseudo-parallel simplification data, easing the Newsela size bottleneck the paper notes for training deep simplification models.
  • The monotone drop from level 1 to level 4 suggests that ROUGE-L captures surface-level rewriting; deeper simplification changes vocabulary and syntax in ways that a lexical overlap metric cannot see, so the 0.566 mean should not be read as semantic equivalence.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The manuscript examines whether an abstractive summarization system can approximate gold-standard text simplification. Using the Newsela corpus, the authors run two BRIO-based summarization protocols (whole-document and paragraph-by-paragraph) over the original articles and compare the outputs to the four levels of professional simplification with ROUGE-L, precision, and recall. They report that paragraph-by-paragraph summarization reaches ROUGE-L 0.654 at level 1 and 0.566 on average, far above document-wide summarization (0.109–0.141), and conclude in Section 5 that although summarization is not a substitute for simplification, it is "similar enough to be used as a baseline to perform simplification on."

Significance. If the finding is robust, the paper provides a useful practical data point: an off-the-shelf abstractive summarizer may produce output with substantial lexical overlap to professional simplifications, making it a candidate starting point for simplification pipelines. The study is transparent in its use of an external benchmark (Newsela) and an external model (BRIO), fits no parameters, and does not rely on circular self-evaluation. The level-wise results are internally consistent, and the paragraph-versus-document contrast is clear. However, the absence of source-text baselines, statistical tests, and released artifacts currently prevents the central claim from being verified; the contribution is descriptive rather than conclusive.

major comments (4)
  1. [Section 3, Table 1] The central number, 0.654, is not interpretable without a source-text baseline. Since the Newsela simplifications and the BRIO summaries are both derived from the same original article, their ROUGE-L overlap may be driven by shared source wording rather than by any simplification-specific behavior. The authors should report ROUGE-L between the original article and each simplification level, and ideally a length-matched extractive baseline (e.g., leading sentences or random paragraphs), over the same 1,913 documents. This control is the minimum needed to support the Section 5 conclusion that the output is "similar enough to be used as a baseline."
  2. [Section 4, Table 1] All results are reported as means over the corpus, with no standard deviations, confidence intervals, or paired significance tests. The document-versus-paragraph gap and the level-wise trends need paired testing (e.g., bootstrap or Wilcoxon signed-rank) to rule out chance and to quantify the spread across documents. Without such tests, the comparative claim that paragraph-by-paragraph summarization is superior is not statistically supported.
  3. [Sections 3 and 5] The claim that paragraph-level summarization can serve as a baseline for simplification is not operationalized. A baseline should be evaluated relative to alternatives such as the original text, an extractive baseline, or an existing simplification system, using a metric like SARI or a human annotation task. The paper's own caveat in Section 5 that ROUGE-L cannot detect semantic equivalence further weakens the leap from a lexical overlap score to a statement about baseline suitability; either weaken the conclusion to a descriptive observation or add the missing comparisons.
  4. [Section 3] The experimental setup is not reproducible as written. The exact BRIO checkpoint, decoding hyperparameters (max length, beam size, length penalty), Newsela version and split, and paragraph segmentation rules are not specified, and no code, model outputs, or evaluation scripts are provided. Because the headline numbers are the sole evidence for the main claim, these details are load-bearing.
minor comments (6)
  1. [Sections 1, 2, and 3] The Newsela corpus size is inconsistent: Sections 1 and 3 state 1,913 original articles, while Section 2 states 1,130 articles; please reconcile the corpus version and counts.
  2. [Section 3] The metric is described as based on "Long Common Sequence"; the correct term is "Longest Common Subsequence" (LCS), and the discussion in Section 4 about "longest common sentences" should use the same terminology.
  3. [Section 1] The phrase "average performance difference of 0.444" is ambiguous; specify that it is the difference between the mean ROUGE-L of the two summarization methods.
  4. [Figures 1 and 2] The figures are referenced but are not self-contained in the current text; ensure that axis labels, legends, and abbreviations are readable and explained in the captions.
  5. [References] Several references contain corrupted diacritics (e.g., "ˇStajner" and "Truic˘a"); please use consistent Unicode or LaTeX escaping.
  6. [Section 3] The statement that BRIO is based on both BART and PEGASUS is confusing because only the BART-based model is used; clarify which checkpoint was selected and why.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the comparison is a direct empirical benchmark between an external summarizer and an external gold-standard simplification corpus, with no fitted parameters or self-citation chains.

full rationale

The paper makes no derivation that reduces to its own inputs. It applies a pretrained external summarizer (BRIO) to Newsela source articles and reports ROUGE-L F1 scores between the generated summaries and the corpus's manual simplifications. No parameter is fitted to Newsela, no simplification references are used to train or fine-tune the model, and no load-bearing claim rests on a self-citation by the authors. The central interpretive step, treating lexical overlap as evidence that summarization approximates simplification, is a validity limitation that the authors themselves acknowledge in Section 5 when they note that ROUGE-L cannot capture semantically correct but lexically different output. That limitation concerns what the metric means, not whether the result is equivalent to its own inputs by construction. The absence of a source-text or length-matched baseline is an experimental design concern that weakens interpretability, but it is not a circularity: the reported numbers are measured outcomes, not consequences of the paper's definitions. Accordingly, no circular step is present and the score is 0.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No numeric free parameters and no invented entities are introduced; the comparison is built from external resources such as Newsela, BRIO, and ROUGE-L. The axioms capture the domain assumptions about gold standards, metric validity, document construction, and model transfer.

assumptions (4)
  • domain assumption Newsela professionally rewritten simplifications are treated as the gold standard for simplification.
    Section 2 and Section 3 use the four labeled levels as the reference outputs for all ROUGE-L comparisons.
  • domain assumption ROUGE-L overlap is a meaningful measure of how closely summarization approximates simplification.
    Section 3 selects ROUGE-L as the evaluation score; Section 5 acknowledges it cannot capture semantic similarity, limiting the inference.
  • domain assumption Rejoined paragraph-level summary outputs can be compared as whole documents to the simplified documents.
    Section 3 describes rejoining paragraph summaries into a single document before scoring.
  • domain assumption A summarizer trained on CNN/Daily Mail and XSum transfers to Newsela news articles without domain adaptation.
    Section 3 chooses BRIO based on availability and performance, with no fine-tuning on Newsela described.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Can summarization approximate simplification? A gold standard comparison." pith.science (2026). https://pith.science/paper/QPGEWORW

@misc{pith2026250116181,
  author       = {Pith},
  title        = {Pith review of: Can summarization approximate simplification? A gold standard comparison},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QPGEWORW}},
  note         = {Machine review of arXiv:2501.16181}
}
read the original abstract

This study explores the overlap between text summarization and simplification outputs. While summarization evaluation methods are streamlined, simplification lacks cohesion, prompting the question: how closely can abstractive summarization resemble gold-standard simplification? We address this by applying two BART-based BRIO summarization methods to the Newsela corpus, comparing outputs with manually annotated simplifications and achieving a top ROUGE-L score of 0.654. This provides insight into where summarization and simplification outputs converge and differ.

Figures

Figures reproduced from arXiv: 2501.16181 by the authors.

Figure 2
Figure 2. The data corresponds to the document￾wide summarization method on the left side and the paragraph-by-paragraph method on the right. When comparing the results from the two pro￾cesses, the overall difference in balance between precision/recall for the document-wide summa￾rization method is immediately noticeable. Even considering the progressive improvement of the precision rate and the lowering of the recall score, … view at source ↗
Figure 1
Figure 1. Representation of the processing pipeline for each article, showing the document-wide method [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Comparison between the different levels of simplified text (1 to 4, left to right) and the two [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 26 canonical work pages

  1. [1]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRINGS urlintro eprinturl eprintpr...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    David Allen. 2009. https://doi.org/10.1016/j.system.2009.09.004 A study of the role of relative clauses in the simplification of news texts for learners of english . System, 37:585--599

    work page doi:10.1016/j.system.2009.09.004 2009

  4. [4]

    Ping Chen, John Rochford, David Kennedy, Soussan Djamasbi, Peter Fay, and Will Scott. 2017. https://doi.org/10.1142/9789813206823_0091 Automatic text simplification for people with intellectual disabilities . In Artificial Intelligence Science and Technology, pages 725--731

    work page doi:10.1142/9789813206823_0091 2017

  5. [5]

    Mahak Gambhir and Vishal Gupta. 2017. https://doi.org/10.1007/s10462-016-9475-9 Recent automatic text summarization techniques: a survey . Artificial Intelligence Review, 47

    doi:10.1007/s10462-016-9475-9 2017

  6. [6]

    Natalia Grabar and Horacio Saggion. 2022. https://hal.science/hal-03701485 Evaluation of Automatic Text Simplification: Where are we now, where should we go from here . In Traitement Automatique des Langues Naturelles , pages 453--463, Avignon, France. ATALA

    work page 2022

  7. [7]

    Max Grusky. 2023. https://doi.org/10.18653/v1/2023.acl-long.107 Rogue scores . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1914--1934, Toronto, Canada. Association for Computational Linguistics

    work page doi:10.18653/v1/2023.acl-long.107 2023

  8. [8]

    Vishal Gupta and Gurpreet Lehal. 2010. https://doi.org/10.4304/jetwi.2.3.258-268 A survey of text summarization extractive techniques . Journal of Emerging Technologies in Web Intelligence, 2

    work page doi:10.4304/jetwi.2.3.258-268 2010

Show all 42 references
  1. [9]

    Karl Moritz Hermann, Tom \'a s Kocisk \'y , Edward Grefenstette, Lasse Espeholt, Will Kay, Mustafa Suleyman, and Phil Blunsom. 2015. https://api.semanticscholar.org/CorpusID:6203757 Teaching machines to read and comprehend . ArXiv, abs/1506.03340

    2015 arXiv

  2. [10]

    Colby Horn, Cathryn Manduca, and David Kauchak. 2014. https://doi.org/10.3115/v1/P14-2075 Learning a lexical simplifier using W ikipedia . In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 458--463, Balti...

    2014 doi

  3. [11]

    Izakian and M

    Z. Izakian and M. Mesgari. 2015. https://doi.org/10.5829/idosi.JAIDM.2015.03.01.05 Fuzzy clustering of time series data: A particle swarm optimization approach . Journal of AI and Data Mining, 3(1):39--46

    2015 doi

  4. [12]

    Bilel Jarraya and Abdelfatteh Bouri. 2012. Metaheuristic optimization backgrounds: A literature review. International Journal of Contemporary Business Studies, 3:31--44

    2012

  5. [13]

    Karel Je ek and Josef Steinberger. 2008. https://api.semanticscholar.org/CorpusID:207742 Automatic text summarization (the state of the art 2007 and new challenges)

    2008

  6. [14]

    Julian Kupiec, Jan Pedersen, and Francine Chen. 1995. https://doi.org/10.1145/215206.215333 A trainable document summarizer . In Proceedings of the 18th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR '95, page 68–73, New Y...

    1995

  7. [15]

    Levenshtein

    Vladimir I. Levenshtein. 1965. https://api.semanticscholar.org/CorpusID:60827152 Binary codes capable of correcting deletions, insertions, and reversals . Soviet physics. Doklady, 10:707--710

    1965

  8. [16]

    Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Veselin Stoyanov, and Luke Zettlemoyer. 2020. https://doi.org/10.18653/v1/2020.acl-main.703 BART : Denoising sequence-to-sequence pre-training for natural language generation, translatio...

    2020 doi

  9. [17]

    Chin-Yew Lin and FJ Och. 2004 a . Looking for a few good metrics: Rouge and its evaluation. In Ntcir workshop

    2004

  10. [18]

    Chin-Yew Lin and Franz Josef Och. 2004 b . https://doi.org/10.3115/1218955.1219032 Automatic evaluation of machine translation quality using longest common subsequence and skip-bigram statistics . In Proceedings of the 42nd Annual Meeting of the Association for Computational L...

    2004

  11. [19]

    Yixin Liu, Pengfei Liu, Dragomir Radev, and Graham Neubig. 2022. http://arxiv.org/abs/2203.16804 Brio: Bringing order to abstractive summarization

    2022 arXiv

  12. [20]

    Hamza Shabbir Moiyadi, Harsh Desai, Dhairya Pawar, Geet Agrawal, and Nilesh M. Patil. 2016. Nlp based text summarization using semantic analysis. International Journal of Advanced Engineering, Management and Science, 2(10)

    2016

  13. [21]

    Cohen, and Mirella Lapata

    Shashi Narayan, Shay B. Cohen, and Mirella Lapata. 2018. https://doi.org/10.18653/v1/D18-1206 Don ' t give me the details, just the summary! topic-aware convolutional neural networks for extreme summarization . In Proceedings of the 2018 Conference on Empirical Methods in Natu...

    2018 doi

  14. [22]

    Nazari and M

    N. Nazari and M. A. Mahdavi. 2019. https://doi.org/10.22044/jadm.2018.6139.1726 A survey on automatic text summarization . Journal of AI and Data Mining, 7(1):121--135

    2019

  15. [23]

    Gustavo Paetzold and Lucia Specia. 2016. https://aclanthology.org/L16-1491 Benchmarking lexical simplification systems . In Proceedings of the Tenth International Conference on Language Resources and Evaluation ( LREC '16) , pages 3074--3080, Portoro z , Slovenia. European Lan...

    2016

  16. [24]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. https://api.semanticscholar.org/CorpusID:11080756 Bleu: a method for automatic evaluation of machine translation . In Annual Meeting of the Association for Computational Linguistics

    2002

  17. [25]

    David Pellow and Maxine Eskenazi. 2014. https://doi.org/10.3115/v1/W14-1210 An open corpus of everyday documents for simplification tasks . In Proceedings of the 3rd Workshop on Predicting and Improving Text Readability for Target Reader Populations ( PITR ) , pages 84--93, Go...

    2014 doi

  18. [26]

    Luz Rello, Ricardo Baeza-Yates, and Horacio Saggion. 2013. https://doi.org/10.1007/978-3-642-37256-8 \_ 41 The impact of lexical simplification by verbal paraphrases for people with and without dyslexia . In Proceedings of the 14th International Conference on Computational Lin...

    2013 doi

  19. [27]

    Advaith Siddharthan. 2014. https://doi.org/10.1075/itl.165.2.06sid A survey of research on text simplification . ITL - International Journal of Applied Linguistics, 165:259--298

    2014 doi

  20. [28]

    Matthew Snover, Nitin Madnani, Bonnie Dorr, and Richard Schwartz. 2009. https://doi.org/10.1007/s10590-009-9062-9 Ter-plus: paraphrase, semantic, and alignment enhancements to translation edit rate . Machine Translation, 23:117--127

    2009 doi

  21. [29]

    Sanja S tajner and Maja Popovic. 2016. https://aclanthology.org/W16-3411 Can text simplification help machine translation? In Proceedings of the 19th Annual Conference of the E uropean Association for Machine Translation , pages 230--242

    2016

  22. [30]

    Sara Stymne, J \"o rg Tiedemann, Christian Hardmeier, and Joakim Nivre. 2013. https://aclanthology.org/W13-5634/ Statistical machine translation with readability constraints . In Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013), pages 375-...

    2013

  23. [31]

    Ilya Sutskever, Oriol Vinyals, and Quoc V. Le. 2014. http://arxiv.org/abs/1409.3215 Sequence to sequence learning with neural networks

    2014 arXiv

  24. [32]

    Ciprian-Octavian Truic a , Andrei-Ionu T Stan, and Elena-Simona Apostol. 2023. https://doi.org/10.1007/s00521-022-07905-y Simplex: a lexical text simplification architecture . Neural Computing and Applications, 35(8):6265--6280

    2023 doi

  25. [33]

    Hoang Van, Zheng Tang, and Mihai Surdeanu. 2021. http://arxiv.org/abs/2109.04604 How may I help you? using neural text simplification to improve downstream NLP tasks . CoRR, abs/2109.04604

    2021 arXiv

  26. [34]

    Tu Thanh Vu, Giang Binh Tran, and Son Bao Pham. 2014. Learning to simplify children stories with limited data. In Intelligent Information and Database Systems, pages 31--41, Cham. Springer International Publishing

    2014

  27. [35]

    Willian Watanabe, Arnaldo Junior, Vinícius Uzêda, Renata Fortes, Thiago Pardo, and Sandra Aluisio. 2009. https://doi.org/10.1145/1621995.1622002 Facilita: Reading assistance for low-literacy readers . pages 29--36

    2009

  28. [36]

    Chih-Hsuan Wei, Robert Leaman, and Zhiyong lu. 2014. https://doi.org/10.1145/2649387.2649420 Simconcept: A hybrid approach for simplifying composite named entities in biomedical text . volume 19

    2014

  29. [37]

    Wei Xu, Chris Callison-Burch, and Courtney Napoles. 2015. https://doi.org/10.1162/tacl_a_00139 Problems in current text simplification research: New data can help . Transactions of the Association for Computational Linguistics, 3:283--297

    2015 doi

  30. [38]

    Wei Xu, Courtney Napoles, Ellie Pavlick, Quanze Chen, and Chris Callison-Burch. 2016. https://doi.org/10.1162/tacl_a_00107 Optimizing statistical machine translation for text simplification . Transactions of the Association for Computational Linguistics, 4:401--415

    2016 doi

  31. [39]

    Jingqing Zhang, Yao Zhao, Mohammad Saleh, and Peter Liu. 2020. https://proceedings.mlr.press/v119/zhang20ae.html PEGASUS : Pre-training with extracted gap-sentences for abstractive summarization . In Proceedings of the 37th International Conference on Machine Learning, volume ...

    2020

  32. [40]

    Ming Zhang, Chengzhang Li, Meilin Wan, Xuejun Zhang, and Qingwei Zhao. 2024. https://doi.org/https://doi.org/10.1016/j.eswa.2023.121364 Rouge-sem: Better evaluation of summarization using rouge combined with semantics . Expert Systems with Applications, 237:121364

    2024

  33. [41]

    Xingxing Zhang and Mirella Lapata. 2017. https://doi.org/10.18653/v1/D17-1062 Sentence simplification with deep reinforcement learning . In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 584--594, Copenhagen, Denmark. Association ...

    2017 doi

  34. [42]

    Zhemin Zhu, Delphine Bernhard, and Iryna Gurevych. 2010. https://aclanthology.org/C10-1152 A monolingual tree-based translation model for sentence simplification . In Proceedings of the 23rd International Conference on Computational Linguistics (Coling 2010), pages 1353--1361,...

    2010

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.