Pith. sign in

REVIEW 4 major objections 6 minor 31 references

New Evaluation Paradigm for Lexical Simplification

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 The paper proposes a sentence-level all-in-one dataset for lexical simplification and shows that a multi-LLM collaboration with majority voting outperforms both pipeline baselines and single-prompt LLMs.

desk verdict A useful new evaluation idea for lexical simplification is undermined by a gold standard contaminated by the evaluated models and a 'significant' claim with no statistics. read the letter →

arxiv 2501.15268 v1 pith:E67TJKPL submitted 2025-01-25 cs.CL

Jipeng Qiang , Minjiang Huang , Yi Zhu , Yunhao Yuan , Chaowei Zhang , Xiaoye Ouyang This is my paper · ORCID
classification cs.CL
keywords lexicalsimplificationevaluationdatasetlargelanguagemodelsmulti-LLMcollaborationcomplexwordidentificationsubstitutegenerationandrankingtextmajorityvoting
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 proposes a new evaluation paradigm for lexical simplification: instead of scoring each complex word separately, judge the whole sentence by annotating every simplifiable complex word and its acceptable substitutes in one dataset. It shows that large language models can produce simplified sentences with a single prompt, bypassing the traditional three-step pipeline of complex word identification, substitute generation, and ranking. To make such outputs measurable, the authors build a 400-sentence all-in-one dataset by combining CWI 2018 data with machine-generated substitute candidates that human annotators vet and extend. They also introduce CoLLS, a multi-LLM framework that runs identification, generation, and sentence-level validation with majority voting, and report that it outperforms both LSBert and single-prompt LLM baselines.

What carries the argument

The key machinery is the all-in-one LS dataset, in which each sentence carries a complete list of complex words and a set of simpler substitutes per word, together with the CoLLS collaboration protocol that decomposes simplification into Complex Word Identification, Substitute Generation, and Validation steps, each executed by multiple LLMs with majority voting. Substitutes are scored by a combination formula that re-ranks outputs from LSBert, GPT-3.5, and Gemini, then human annotators judge simplicity and meaning preservation for each candidate and may add their own. The difficulty-weighted F1-20 metric weights each simplification by how many of the 20 original annotators marked the word as complex.

What would settle it

Re-annotate a random sample of the dataset using only human-generated substitutes and human-only judgments, then rerun the Table 4 comparisons; if LSBert or single-prompt methods match or beat CoLLS on this human-only gold standard, the reported superiority would be an artifact of the LLM-influenced candidate pool rather than a general result.

Watch

Extended reading notes

Core claim

The central discovery is that a single dataset containing all complex words in each sentence and their substitutes, built by human-machine collaboration, supports direct end-to-end evaluation of sentence-level lexical simplification, and that on this dataset LLM-based methods outperform pipeline-based small-model baselines like LSBert. The paper further claims that multi-LLM collaboration—where one LLM proposes complex words, another generates substitutes, and a third validates the simplified sentence, each with majority voting—significantly outperforms single-prompt LLM methods and existing baselines on both unweighted F1 and the difficulty-weighted F1-20 metric.

Load-bearing premise

The evaluation gold standard is fair even though it was built from substitutes suggested by the same families of LLMs that are later scored against it, and by LLM-assisted annotations that humans may not have fully corrected.

Editorial extensions

If this is right

  • Simplification models can be compared end to end on whole sentences rather than word by word, making single-prompt LLM outputs directly measurable.
  • LLM-based simplification no longer needs to be staged as separate CWI/SG/SR models, since a single prompt already beats a small-model pipeline.
  • The majority-voting validation step is load-bearing: removing it lowers F1-20, so consensus-based sentence-level checking appears to improve output quality.
  • Difficulty weighting changes rankings by rewarding models that simplify genuinely hard words, so evaluation should report both F1 and F1-20.
  • The annotation method extends to other languages by starting from existing CWI annotations and adding machine-generated candidates for humans to vet.

Reading between the lines

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

  • One implicit consequence is that the pseudo-substitute pool's composition, drawn from LSBert, GPT-3.5, and Gemini, may favor methods using the same candidate-generating families; a cleaner test would compare against a pool built from fully human suggestions only.
  • If CoLLS-style validation is applied to other sentence-generation tasks, the same propose-generate-verify decomposition could become a generic template for LLM text simplification and controlled text revision.
  • The F1-20 difficulty-weighting idea could be reused in readability or accessibility evaluations wherever annotator agreement counts are available.
  • A testable extension is to expand the 400-instance dataset to more genres and languages and check whether the ranking of methods remains stable.
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 paper proposes a new evaluation paradigm for lexical simplification (LS) based on an 'all-in-one' dataset in which every complex word in a sentence is annotated with a set of suitable simpler substitutes. The dataset is constructed by generating a candidate substitute pool with LSBert, GPT-3.5, and Gemini, fusing the candidates via Eq. (1)–(2), and then having human annotators judge the candidates with the aid of LLM predictions that the same models provide. The paper evaluates LSBert, GPT-3.5, Llama-3, and a proposed multi-LLM collaboration framework CoLLS on this dataset using F1 and a difficulty-weighted F1-20. The main claimed result is that CoLLS significantly outperforms existing baselines and single-prompt LLM methods. The paper also presents an ablation study on the number of LLMs in CoLLS.

Significance. If the dataset and evaluation are sound, the paper would contribute a useful resource and a new end-to-end way to evaluate LS, and it demonstrates that LLM-based single-prompt simplification is competitive and that a multi-LLM collaboration can further improve results. The human-machine annotation workflow is pragmatic, and the explicit checks of precision (94%) and coverage (95%) are good practices. However, because the gold standard is partly constructed from the outputs of the very models being evaluated, and because the headline comparisons lack statistical support, the empirical claims are currently not reliable enough to support the conclusions.

major comments (4)
  1. [§3.3–§3.4, Eq. (1)–(2), Table 4] The gold standard is endogenous to the evaluated systems. The pseudo-substitute pool is generated by LSBert, GPT-3.5, and Gemini and fused to a top-12 list via Eq. (1)–(2), and in §3.4 the same GPT-3.5 and Gemini pre-annotate each candidate, with their recommendations displayed to human annotators on the annotation website. Table 4 then scores LSBert, GPT-3.5, and CoLLS (built on GPT-3.5) against this gold. As a result, GPT-3.5-generated substitutes are disproportionately already in the pool and pre-endorsed by GPT-3.5 itself, while valid LSBert substitutes that fell outside the top-12 fusion are never shown to annotators and can only enter the gold if an annotator spontaneously adds them. The reported 95% coverage in §3.5 does not remove this asymmetry: the two new human evaluators produced substitutes after seeing the same task and may share the pool's lexical biases, and the 17 uncovered substitutes are exactly the ones penalized. Consequently, the claim in §5.2 that CoLLS 'significantly outperforms' other methods rests on a benchmark whose construction privileges the models being scored; this needs to be addressed, for example by a blind, independent re-annotation of a held-out subset or by re-building the pool without the evaluated models.
  2. [§5.1–§5.2, Table 4] No statistical support is given for the central comparative claim. The results appear to come from a single run of each method; no error bars, standard deviations, or significance tests are reported. On a dataset of only 400 instances, the F1 differences between CoLLS(GPT-3.5) and the best single-prompt methods are in the 0.06–0.08 range, so the abstract's 'significantly outperforms' is unsupported. The paper should either report multiple runs with variance estimates and a paired test, or soften the claim to a descriptive comparison.
  3. [§3.2, §5.1] The post hoc removal of complex words without suitable substitutes changes the evaluation target relative to the original CWI-2018 labels and may bias the comparison against LSBert. The paper states that 'if a word does not have a suitable substitute, it cannot be considered a complex word' and filters such words out of the dataset, but LSBert was trained on the original CWI labels, where these words were annotated as complex. Moreover, the 'suitability' determination was made through the same human-machine loop that produced the gold, so the filtering is not independent of the gold construction. The authors should quantify how often filtering occurred and, ideally, run an additional evaluation on the unfiltered word set with a 'no simplification needed' option.
  4. [§5.1 (Baselines)] The evaluation of LSBert is underspecified. LSBert is a pipeline with a CWI classifier and a substitute generator/ranker, but the paper does not describe how it produces simplified sentences or how its outputs are aligned with the whole-sentence gold. It is unclear whether LSBert's CWI predictions are used to select which words to simplify, how the substitute set is restricted (e.g., whether it is limited to the gold-annotated words), and how ties and multiple substitutes per word are handled. A mismatch between LSBert's identification criterion and the gold's post hoc filtering can depress recall independently of substitute quality. The authors should specify the exact decoding and scoring procedure for LSBert in Table 4.
minor comments (6)
  1. [§5.4] There are minor typographical errors: 'paer' should be 'paper', and 'a all-in-one' should be 'an all-in-one'.
  2. [§5.1] 'AgentLS' is mentioned in the Baselines paragraph but is never defined, described, or evaluated in Table 4; this appears to be a leftover from a previous draft.
  3. [Table 3] The header '3' and '4' is cryptic; the voting threshold should be clearly defined (e.g., 'at least 3 of 4 models agree').
  4. [§5.3] The ablation uses 'M' for the majority threshold, while elsewhere 'm' is used; the notation should be consistent.
  5. [General] The dataset is not released or linked; given that the paper's proposal is a new evaluation paradigm, releasing the dataset is essential for reproducibility and for the community to adopt it.
  6. [Appendix A.1] The annotation website is described, but there is no information on the number of annotators, their qualifications, or inter-annotator agreement beyond the consistency test with LLMs; a brief description would help assess the reliability of the gold labels.

Circularity Check

1 steps flagged · score 6.0 of 10

Gold-standard substitutes are generated and pre-annotated by the same GPT-3.5/LSBert systems later evaluated, partially forcing the claimed LLM advantage.

  1. self definitional [Section 3.3 (Eq. 1-2), Section 3.4, Table 4]
    "we employ three different LS methods to generate a set of pseudo substitutes for each complex word. Here, we chose one of the best small model-based method LSBert (Liu et al., 2023) and two LLM-based methods (GPT3.5 and Gemini1.0). ... we continue to offer aid to annotators on the suitability of pseudo substitutes. We choose two LLMs (GPT3.5 and Gemini1.0), combined with few-shot prompting (Direct) and chain-of-thought (COT) prompting strategies, to annotate the pseudo-substitution words."

    The gold set is a human-filtered subset of a candidate pool built from LSBert, GPT-3.5, and Gemini, with GPT-3.5 and Gemini pre-annotations displayed to annotators. The same GPT-3.5 and LSBert are then scored against this gold in Table 4. A substitute produced by GPT-3.5 during evaluation is disproportionately likely to lie in the pool constructed via Eq. (1-2) and therefore in the human-approved gold, while a valid substitute from a method outside the pool (e.g., Llama3) counts as correct only if an annotator spontaneously added it. The reported F1 gap between GPT-3.5/CoLLS and LSBert is thus partly manufactured by the pool's composition, not purely by simplification quality.

full rationale

The central contribution is a new all-in-one LS evaluation paradigm, and the load-bearing claim is that CoLLS and single-prompt LLM methods outperform the LSBert baseline. The dataset construction reuses the evaluated models: Section 3.3 generates pseudo substitutes with LSBert, GPT-3.5, and Gemini, and Section 3.4 has GPT-3.5 and Gemini pre-annotate the candidates that are shown to human annotators. Table 4 then evaluates LSBert, GPT-3.5, Llama3, and CoLLS (built on GPT-3.5 or Llama3) against the resulting gold. This creates a partial circularity: the gold standard is not independent of the systems being measured. The human annotation step adds genuine external judgment, and the reported 95% coverage by two new human evaluators suggests the dataset is broadly representative, so the circularity is not total. However, the human annotators were not blind; they saw LLM recommendations, and Table 3's high human-LLM consistency could reflect anchoring as much as LLM quality. The Limitations section acknowledges the small scale but does not flag the endogenous gold construction, and the dataset is not released, so the magnitude of the bias cannot be externally checked. The word "significantly" in the abstract is also not supported by significance tests, which is a correctness concern rather than a circularity one. On balance, the evaluation paradigm is innovative but the reported superiority of GPT-3.5-based methods is partially baked into the test-set construction, warranting a score of 6 rather than a higher one.

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

The paper's central artifact is a small human-machine-built dataset; the free parameters show how much of the gold is shaped by arbitrary cutoffs, hand-chosen weights, and majority thresholds. No new physical or theoretical entities are introduced.

free parameters (4)
  • Candidate combination weight constants (5 - 0.5*index) = 5, 0.5
    Eq. (1)-(2) use hand-chosen weights to fuse substitute lists from three generators; no tuning or justification is given, and the resulting top-12 pool bounds what annotators can approve.
  • Top-12 candidate cutoff = 12
    Section 3.3 selects top 12 candidates for each complex word; this cutoff directly controls the gold substitute space.
  • CoLLS voting thresholds (N=3, m=2) = N=3, m=2
    Section 5.1 sets every CoLLS step to N=3, m=2; the ablation in Section 5.3 covers only WikiNews and CoLLS(Llama3), so the default is a design choice rather than a tuned optimum.
  • Few-shot demonstration counts = 4 (single-prompt); 2/4/6 (CoLLS steps)
    Section 5.1 chooses four demonstrations for single-prompt baselines and 2, 4, 6 for CoLLS roles; these counts affect outputs and are not justified by experiments.
assumptions (4)
  • domain assumption CWI 2018 complex-word annotations are reliable and complete for identifying simplifiable words.
    Section 3.2 constructs the dataset directly from CWI 2018 annotations without re-validating the set of complex words.
  • ad hoc to paper A word counts as complex for LS only if a suitable simpler substitute exists.
    Section 3.2 removes complex words without substitutes, redefining 'complex word' away from the CWI definition.
  • domain assumption The two annotation criteria (simpler, meaning-preserving) are sufficient to judge substitute quality.
    Section 3.4 instructs annotators to use only these two yes/no questions; fluency and register are not separately assessed in gold construction.
  • domain assumption LLM pre-annotations guide human annotators without biasing the gold labels.
    Section 3.5 reports human-LLM consistency, but high agreement may simply mean annotators followed the LLM recommendations shown on the annotation website.

how reviews work

0 comments
Cite this review

Pith. "Pith review of New Evaluation Paradigm for Lexical Simplification." pith.science (2026). https://pith.science/paper/E67TJKPL

@misc{pith2026250115268,
  author       = {Pith},
  title        = {Pith review of: New Evaluation Paradigm for Lexical Simplification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E67TJKPL}},
  note         = {Machine review of arXiv:2501.15268}
}
read the original abstract

Lexical Simplification (LS) methods use a three-step pipeline: complex word identification, substitute generation, and substitute ranking, each with separate evaluation datasets. We found large language models (LLMs) can simplify sentences directly with a single prompt, bypassing the traditional pipeline. However, existing LS datasets are not suitable for evaluating these LLM-generated simplified sentences, as they focus on providing substitutes for single complex words without identifying all complex words in a sentence. To address this gap, we propose a new annotation method for constructing an all-in-one LS dataset through human-machine collaboration. Automated methods generate a pool of potential substitutes, which human annotators then assess, suggesting additional alternatives as needed. Additionally, we explore LLM-based methods with single prompts, in-context learning, and chain-of-thought techniques. We introduce a multi-LLMs collaboration approach to simulate each step of the LS task. Experimental results demonstrate that LS based on multi-LLMs approaches significantly outperforms existing baselines.

Figures

Figures reproduced from arXiv: 2501.15268 by the authors.

Figure 1
Figure 1. Flowchart of current LS methods and LLM [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An overview of the methodology of the LS [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Prompt template for one-step LS. accuracy of the substitutions made in the given sen￾tence and complex word pairs. A total of 150 in￾stances were randomly selected, with 50 instances chosen from one of three text genres. A new an￾notator, proficient in the English language, was assigned the task of assessing the precision of the substitutions within the selected instances. The precision of the substitutions was comp… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Prompt template for LLM-based LS method (COT). This technique aims to enhance the results by lever￾aging semantically similar examples (few-shot) or utilizing uncertainty and diversity for demonstra￾tion refinement and evaluation. The few-shot-based LS prompt is shown …
Figure 5
Figure 5. Figure 5: The framework of CoLLS. Each step defines the roles and tasks of LLMs. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Effect of varying the number of LLMs (N) for CoLLS(Llama3) on WikiNews. (a) the results on metric F1, and (b) the results on metric F1-20. the F1 value does not change significantly, but the F1-20 value is somewhat lower. This suggests that the Validation step can enha…
Figure 7
Figure 7. Figure 7: Screenshot of an annotation example on the [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Prompt template of Substitute Generator in [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Prompt template (Direct) for Manual annota [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 12
Figure 12. Figure 12: Prompt template of CoLLS (Sentence com￾parator). C Case study Here, we randomly choose 5 instances from WikiNews for analysis in [PITH_FULL_IMAGE:figures/full_fig_p012_12.png]
Figure 10
Figure 10. Figure 10: Prompt template (Direct) for Manual annota [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Prompt template of CoLLS (Complex word analyst). ####Instruction#### You are an editor with a solid writing foundation and exten￾sive reviewing experience, given an original SENTENCE 1, and its simplified SENTENCE 2, compare whether SEN￾TENCE 2 has the same meaning as…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 23 canonical work pages

  1. [1]

    online" 'onlinestring :=

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

  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]

    Dennis Aumiller and Michael Gertz. 2022. U ni HD at TSAR -2022 shared task: Is compute all we need for lexical simplification? In Proceedings of the Workshop on Text Simplification, Accessibility, and Readability (TSAR-2022), pages 251--258

    work page 2022

  4. [4]

    S Bott, L Rello, B Drndarevic, and H. Saggion. 2012. Can spanish be simpler? lexsis: Lexical simplification for spanish. In Proceedings of COLING, pages 357--374

    work page 2012

  5. [5]

    Weize Chen, Yusheng Su, Jingwei Zuo, Cheng Yang, Chenfei Yuan, Chen Qian, Chi-Min Chan, Yujia Qin, Yaxi Lu, Ruobing Xie, et al. 2023. Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors in agents. arXiv preprint arXiv:2308.10848

    arXiv 2023

  6. [6]

    Yao Fu, Hao Peng, Ashish Sabharwal, Peter Clark, and Tushar Khot. 2023. https://arxiv.org/abs/2210.00720 Complexity-based prompting for multi-step reasoning . Preprint, arXiv:2210.00720

    arXiv 2023

  7. [7]

    Goran Glava s and Sanja S tajner. 2015. Simplifying lexical simplification: do we need simplified corpora? In ACL, pages 63--68

    work page 2015

  8. [8]

    Sian Gooding and Ekaterina Kochmar. 2018. Complex word identification with ensemble-based voting. In In Proceedings of the Thirteenth Workshop on Innovative Use of NLP for Building Educational Applications, pages 184--194

    work page 2018

Show all 31 references
  1. [9]

    Sian Gooding and Ekaterina Kochmar. 2019. Complex word identification as a sequence labelling task. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 1148--1153

    2019

  2. [10]

    Kang Liu, Jipeng Qiang, Yun Li, Yunhao Yuan, Yi Zhu, and Kaixun Hua. 2023. Multilingual lexical simplification via paraphrase generation. In 26th European Conference on Artificial Intelligence,, volume 372, pages 1529--1535

    2023

  3. [11]

    Mounica Maddela and Wei Xu. 2018. A word-complexity lexicon and a neural readability ranking model for lexical simplification. In EMNLP, pages 3749--3760

    2018

  4. [12]

    Kai North, Tharindu Ranasinghe, Matthew Shardlow, and Marcos Zampieri. 2024. Multils: A multi-task lexical simplification framework. arXiv preprint arXiv:2402.14972

    2024 arXiv

  5. [13]

    Kai North, Marcos Zampieri, and Matthew Shardlow. 2023. Lexical complexity prediction: An overview. ACM Computing Surveys, 55(9):1--42

    2023

  6. [14]

    Gustavo Paetzold and Lucia Specia. 2017 a . Lexical simplification with neural ranking. In ACL: Volume 2, Short Papers, pages 34--40

    2017

  7. [15]

    Gustavo H Paetzold and Lucia Specia. 2016. Unsupervised lexical simplification for non-native speakers. In AAAI, pages 3761--3767

    2016

  8. [16]

    Gustavo H Paetzold and Lucia Specia. 2017 b . A survey on lexical simplification. In Journal of Artificial Intelligence Research, volume 60, pages 549--593

    2017

  9. [17]

    Ellie Pavlick and Chris Callison-Burch. 2016. Simple ppdb: A paraphrase database for simplification. In ACL: Volume 2, Short Papers, pages 143--148

    2016

  10. [18]

    Jipeng Qiang, Yun Li, Yi Zhu, Yunhao Yuan, Yang Shi, and Xindong Wu. 2021 a . Lsbert: Lexical simplification based on bert. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:3064--3076

    2021

  11. [19]

    Jipeng Qiang, Yun Li, Yi Zhu, Yunhao Yuan, and Xindong Wu. 2020. Lexical simplification with pretrained encoders. Thirty-Fourth AAAI Conference on Artificial Intelligence, page 8649–8656

    2020

  12. [20]

    Jipeng Qiang, Xinyu Lv, Yun Li, Yunhao Yuan, and Xindong Wu. 2021 b . Chinese lexical simplification. IEEE Transactions on Audio, Speech and Language Processing., 29:1819--1828

    2021

  13. [21]

    Horacio Saggion. 2017. Automatic text simplification. Synthesis Lectures on Human Language Technologies, 10(1):1--137

    2017

  14. [22]

    Sandaru Seneviratne and Hanna Suominen. 2024. ANU at MLSP -2024: Prompt-based lexical simplification for E nglish and S inhala. In Proceedings of the 19th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2024), pages 599--604

    2024

  15. [23]

    Matthew Shardlow, Richard Evans, Gustavo Henrique Paetzold, and Marcos Zampieri. 2021. S em E val-2021 task 1: Lexical complexity prediction. In Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021), pages 1--16

    2021

  16. [24]

    Kim Cheng Sheang, Daniel Ferr \'e s, and Horacio Saggion. 2022. Controllable lexical simplification for E nglish. In Proceedings of the Workshop on Text Simplification, Accessibility, and Readability (TSAR-2022), pages 199--206

    2022

  17. [25]

    Kim Cheng Sheang and Horacio Saggion. 2023. Multilingual controllable transformer-based lexical simplification. arXiv preprint arXiv:2307.02120

    2023 arXiv

  18. [26]

    Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou

    Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V Le, Ed H. Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2023. https://openreview.net/forum?id=1PL1NIMMrw Self-consistency improves chain of thought reasoning in language models . In The Eleventh International Conferenc...

    2023

  19. [27]

    Minghao Wu, Yulin Yuan, Gholamreza Haffari, and Longyue Wang. 2024. (perhaps) beyond human translation: Harnessing multi-agent collaboration for translating ultra-long literary texts. arXiv preprint arXiv:2405.11804

    2024 arXiv

  20. [28]

    Sander Wubben, Antal Van Den Bosch, and Emiel Krahmer. 2012. Sentence simplification by monolingual machine translation. In EMNLP, pages 1015--1024

    2012

  21. [29]

    Yang, Damian Dailisan, Marcin Korecki, Carina I

    Joshua C. Yang, Damian Dailisan, Marcin Korecki, Carina I. Hausladen, and Dirk Helbing. 2024. https://arxiv.org/abs/2402.01766 Llm voting: Human choices and ai collective decision making . Preprint, arXiv:2402.01766

    2024 arXiv

  22. [30]

    Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2022. React: Synergizing reasoning and acting in language models. arXiv preprint arXiv:2210.03629

    2022 arXiv

  23. [31]

    Seid Muhie Yimam, Chris Biemann, Shervin Malmasi, Gustavo H Paetzold, Lucia Specia, Sanja S tajner, Ana \" s Tack, and Marcos Zampieri. 2018. A report on the complex word identification shared task 2018. pages 66--78

    2018

Pith tools

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