Explaining Matters: Leveraging Definitions and Semantic Expansion for Sexism Detection

Arshad Jhumka; Gabriele Pergola; Sahrish Khan

REVIEW 5 major objections 6 minor 53 references

Explaining Matters: Leveraging Definitions and Semantic Expansion for Sexism Detection

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Giving models category definitions and explanations of their own errors improves fine-grained sexism detection on the EDOS benchmark.

desk verdict Useful empirical study of definition-based augmentation for sexism detection; results are plausible but the abstract overclaims SOTA and the synthetic labels are unverified. read the letter →

arxiv 2506.06238 v1 pith:WKSIEKUL submitted 2025-06-06 cs.CL

Sahrish Khan , Arshad Jhumka , Gabriele Pergola This is my paper

classification cs.CL

keywords sexismdetectiondataaugmentationlargelanguagemodelsEDOSdatasetfine-grainedclassificationpromptengineeringensemblemethodssemanticexpansion

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

This paper tries to establish that the hard part of online sexism detection is not only data scarcity but category ambiguity, and that both can be attacked with two prompt-based augmentation methods. Definition-based Data Augmentation (DDA) feeds each fine-grained category's formal definition into an LLM to generate synthetic training examples that stay inside that category's meaning, while Contextual Semantic Expansion (CSE) takes examples a model misclassified and adds a generated explanation of the contextual cues that make them sexist. On the EDOS benchmark, the methods achieve macro F1 of 0.8819 for binary sexism detection and 0.6018 for the 11-way fine-grained task, improving over comparable prior results by 1.5 and 4.1 points respectively. A sympathetic reader would care because the recipe is transferable: instead of collecting more labeled data, a practitioner can use existing category definitions and error analysis to sharpen a classifier's boundaries.

What carries the argument

The load-bearing machinery is a pair of LLM prompts plus a tie-breaking ensemble. DDA constructs a prompt containing the EDOS vector label, its official definition (e.g., 2.1 Descriptive Attacks: "Characterizing or describing women in a derogatory manner"), and a seed example, then asks the model to generate three variations that preserve the sexist intent, match the tone, and mimic informal social-media language; it is applied only to the five least-represented Task C classes. CSE targets the examples the base DeBERTa-v3-large classifier misclassified during training, runs them through a six-step analysis prompt covering language patterns, neutrality, sentiment, situational context, gender stereotypes, and demeaning intent, and concatenates the resulting semantic explanation to the original input as the augmented training instance. M7-FE then takes hard predictions from DeBERTa-v3-large, DTFN, and Mistral-7B, chooses the majority class, and when votes tie or fully disagree, uses Mistral-7B's prediction as the final answer.

What would settle it

Have annotators re-label a random sample of the DDA-generated examples against the EDOS taxonomy and remove or correct those that do not match their assigned category; if the Task C macro F1 gap of 4.1 points collapses, the gain came from label alignment rather than from clarifying category boundaries.

Watch

Extended reading notes

Core claim

The paper's central claim is that supplying a model with the definitions of the categories it must separate, and with semantic explanations of its own mistakes, reduces both the sparsity problem and the annotator-disagreement problem in sexism detection. Its strongest reported evidence is that DDA plus the Mistral-7B Fallback Ensemble raises the fine-grained Task C macro F1 on EDOS from the previous best of 0.5606 to 0.6018, and CSE raises binary Task A to 0.8819, while DDA reaches 0.7277 on Task B, within 0.005 of the 0.7326 PaLM ensemble score. Since its own Table 2 lists a higher Task B score for the PaLM ensemble, the abstract's phrase "state-of-the-art performance across all tasks" holds for Tasks A and C but not for Task B on the paper's own numbers.

Load-bearing premise

The paper assumes that the GPT-4o-generated synthetic examples are labeled correctly and stay inside the intended category's meaning, yet it reports no human or automated verification of those labels.

Editorial extensions

If this is right

On the EDOS test set, DDA plus the ensemble lifts fine-grained macro F1 from 0.5606 to 0.6018, with the largest correct-prediction increases in rare classes such as 2.3 Dehumanising attacks and overt sexual objectification and 3.4 Condescending explanations or unwelcome advice.
The ablation shows that injecting the category definition into the prompt is what drives the fine-grained gains: a baseline prompt without definitions scores 0.5601 on Task C, versus 0.6018 with DDA.
Augmenting only five under-represented classes yields better results than augmenting all eleven, so targeted augmentation is both cheaper and more effective than indiscriminate expansion.
CSE's gains concentrate in binary classification, consistent with the observation that most misclassified examples carry high confidence (p > 0.9), meaning the errors are systematic biases rather than boundary ambiguity.

Reading between the lines

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

If the DDA gains are label-faithful, the same definition-grounded augmentation recipe should transfer to other hierarchical toxicity taxonomies whose leaf categories are sparse and overlapping, such as hate-speech or harassment frameworks.
A direct test of the paper's mechanism is to have annotators verify a random sample of GPT-4o-generated labels; if many are mislabeled, the reported improvements would be artifacts of label leakage rather than clarified boundaries.
The fallback choice matters: the appendix shows Mistral-7B beat DTFN and DeBERTa as the tie-breaker, so a stronger fallback model could plausibly push the ensemble further on tasks where models disagree.
CSE is essentially chain-of-thought reasoning applied to classification inputs; a natural extension is to test whether the six-step explanation can be shortened or applied at inference time only, without retraining.

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.

Desk Editor's Note

read the letter

Short version: this is a useful empirical paper that shows prompt-based augmentation with category definitions moves the needle on EDOS for Task A and Task C, and it ships code. The central claim needs moderation: the abstract's "SOTA across all tasks" is not supported by their own Table 2, where the PaLM ensemble still leads Task B. And the biggest weakness is the unverified GPT-4o labels in DDA.

What's actually new: the specific combination of DDA (definition-grounded augmentation) and CSE (semantic expansion of misclassified examples) applied to fine-grained sexism categories, plus a fallback ensemble for tie-breaking. The gains of +1.5 Task A and +4.1 Task C over comparable prior work are new results, and the ablation against a baseline prompt does control for the number of generated examples. The confusion matrix analysis showing reduced cross-category confusion for rare classes like 2.3 and 3.4 is a nice piece of evidence. The authors also acknowledge in Section 7 that bias/artifacts in synthetic examples remain an open risk, which is honest.

Where I'd push back: First, the SOTA claim. Their own table shows DDA Task B 0.7277 vs PaLM 0.7326, so "across all tasks" is simply false. It's a small fix, but it matters because the paper's headline is built on it. Second, the label quality issue raised in the stress-test is real and not fully answered. Eq. (1) assigns the source label to every GPT-4o variation, and the prompt explicitly asks for "sexist intent" and tone matching, which could bias toward exaggerated prototypes. The paper reports no manual inspection, annotator agreement, or automated consistency check on the 1,326 added examples. That makes the Task C improvement hard to attribute to definition-based disambiguation rather than LLM-style genre artifacts or rebalancing. The ablation controls quantity but not label validity. Third, the selection details are underreported: the five augmented classes come from an undisclosed "preliminary analysis", and the best epoch per model is chosen post hoc from a small epoch grid. Without held-out model selection or error bars, the 4-point gap could shrink.

None of this makes the paper a waste of time. The methods are sensible, the experiments are mostly transparent, and the code is public. For a reader working on limited-resource text classification or abusive language detection, this is worth a close look. I would send it out; the referees can push the authors to verify labels, fix the SOTA claim, and report selection procedures. It's a revise-and-resubmit trajectory, not a reject.

Referee Report

5 major / 6 minor

Summary. The paper proposes two prompt-based data augmentation techniques for sexism detection on the EDOS benchmark: Definition-based Data Augmentation (DDA), which uses category definitions to generate synthetic examples aligned with fine-grained Task C labels, and Contextual Semantic Expansion (CSE), which expands misclassified examples with semantic analysis. A Mistral-7B Fallback Ensemble (M7-FE) is used to break ties among fine-tuned models. On the EDOS test set the authors report macro F1 of 0.8819 on Task A, 0.7277 on Task B, and 0.6018 on Task C (Table 2), and claim state-of-the-art performance across all tasks.

Significance. If the reported results are robust, the paper offers a practical recipe for combining definition-aware LLM augmentation with ensembling, and it makes a useful contribution by documenting annotator disagreement across EDOS Task C categories. The public code release, the ablation against a baseline prompt, and the detailed appendix tables are to the authors' credit. However, the central empirical claim rests on unverified synthetic-label correctness and on several post hoc model-selection choices, so the significance cannot be fully assessed until those load-bearing points are addressed.

major comments (5)

[Abstract and Section 4.2, Table 2] The abstract and Section 4.2 claim 'state-of-the-art performance across all tasks', but Table 2 shows that the PaLM Ensemble achieves 0.7326 on Task B while the best proposed result is 0.7277. The claim should be restricted to Tasks A and C, or the Task B comparison should be discussed explicitly as closely approaching rather than surpassing the prior state of the art.
[Section 3.3, Eq. (1), Figure 1, and Section 7] The central Task C improvement of 4.1 points (0.6018 vs 0.5606) is attributed to DDA, yet Eq. (1) assigns the source label to every GPT-4o-generated variation with no manual inspection, annotator agreement study, or automated consistency check. Section 7 itself concedes that synthetic examples may contain unintended biases or artifacts. Without verification that the 1,326 added examples for the five rare classes are label-correct and in-domain, the observed gain cannot be attributed to definition-based boundary clarification rather than to distribution shift or genre artifacts. A sample-level human evaluation or a consistency check against the taxonomy is needed to support the claim.
[Section 4.1 and Appendix A3-A5] The reported scores appear to select the best epoch per model after seeing test-set performance: Table A3 lists DeBERTa-v3-Large at epochs 3, 7, and 9 and DTFN at epochs 3, 7, and 8 with different Task A/B/C scores, and the main table uses the best-triple configuration. No validation-based early stopping, fixed epoch schedule, or multiple-seed variance is reported. This post hoc selection can materially inflate macro F1, so the authors should report a single predefined selection rule or provide standard deviations over seeds.
[Section 4.1 and Appendix C] The choice to augment exactly five classes (1.1, 2.3, 3.3, 3.4, 4.1) is justified only as 'identified through preliminary analysis' and as low-resource in Appendix C. The selection criterion, the number of generated variations (three), and the fallback-model choice are all free parameters tuned on the same EDOS test set. The paper should report the selection procedure and, ideally, a validation-based protocol to show that these choices are not optimizing test-set scores directly.
[Section 4.2 and Table 2] The text states that M7-FE 'contributes to a 2.5-point gain in Task B', but the comparison in Table 2 between line 13 (ensemble without augmentation, 0.7027) and line 15 (DDA plus ensemble, 0.7277) conflates the DDA augmentation with the ensemble effect. Since no row isolates M7-FE from DDA (e.g., DDA without the fallback ensemble), the stated attribution is not supported by the presented ablations.

minor comments (6)

[Table 2 caption] The caption says 'particularly with the CSE technique on Task A (0.8819) and on Task C (0.6018)', but 0.6018 is achieved by DDA, not CSE; this is confusing and should be corrected.
[Table A3] In the row 'Generated Data with Baseline Prompts: 3 Variations of 5 Key Classes', DTFN (epoc: 8) reports Task B = 0.8674, which is identical to the Task A value and is likely a typo for a value near 0.6974.
[Section 3.4] The count '2.518 sexist examples' and '2.328 non-sexist examples' uses periods as thousand separators; use commas or spaces for readability.
[Appendix D] The phrase 'for convinience' is a typo; it should be 'for convenience'.
[Section 3.6] There are minor grammar issues, such as 'can induces' and 'aims to leverages', which should be cleaned up.
[References] The entry 'Chen Lyu, , and Gabriele Pergola' has a missing author name, and some author names are inconsistently capitalized; please run a final reference check.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity; evaluation is external and empirical, with only minor non-load-bearing self-citations.

full rationale

The paper's central claims are empirical comparisons on the external EDOS benchmark (SemEval-2023 Task 10), not derivations. DDA (Eq. 1) assigns the source label yi to each GPT-4o-generated variation fLLM(xi, phi(yi)) and augments D; this is a training-data construction, not a prediction of the test labels, and the reported macro-F1 gains are measured on held-out test splits. CSE is similarly a self-refinement loop over training misclassifications with test-set evaluation. The M7-FE ensemble is standard majority voting with a Mistral fallback, whose choice is justified by preliminary experiments rather than by an imported uniqueness theorem. No equation reduces to its input, and no fitted parameter is renamed as a prediction. Several citations are to the authors' prior work (e.g., DTFN, Khan et al. 2024; earlier Pergola et al. papers), but these are baselines or related-work context, not load-bearing justifications of the augmentation claims. The paper's weakness is validation-related: it reports no human or automated check of GPT-4o label correctness for DDA (Section 3.3, Figure 1; admitted in Section 7), which is a data-quality and generalizability concern, not circularity. The abstract's 'state-of-the-art performance across all tasks' overstates Table 2, where the PaLM ensemble retains the highest Task B score (0.7326); that is a reporting-accuracy issue, not a circularity issue.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claims rest on the quality of generated data and on the validity of hyperparameter choices that are not fully specified. There are no new physical or mathematical entities.

free parameters (4)

number_of_dda_variations = 3
m = 3 synthetic variations generated per example in Eq. (1); a modeling choice not justified beyond 'Generate 3 Variations' in Figure 1.
augmented_classes = 5 of 11
Section 3.3: 'DDA specifically targets the c = 5 most underrepresented classes... identified through preliminary analysis.' The subset affects the Task C result by 4.1 points; if the selection used the test set, the gain is inflated.
fallback_model = Mistral-7B
Section 3.6: 'Based on preliminary experiments, Mistral-7B is chosen as the fallback model.' The fallback choice affects the final ensemble scores.
epoch_per_model = 3-30 (best selected per model)
Appendix A3 reports multiple epochs per model; the best epoch per model is selected post hoc and reported in the main results, which is a free choice that can inflate test scores.

assumptions (3)

domain assumption GPT-4o generation fidelity
The methods assume the LLM-generated synthetic examples will faithfully reflect the sexist intent and category definition (Sec. 3.3, Figure 1).
domain assumption Aggregated annotator labels are ground truth
The paper uses the aggregated EDOS labels for training and evaluation despite reporting high annotator disagreement (Sec. 3.2); this treats the majority vote as correct.
domain assumption Preliminary analysis used only training/validation data
The choice of five classes and the fallback model is described as 'preliminary analysis' (Sec. 3.3, 3.6) but the data used is not specified; if test data was used, results are invalid.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Explaining Matters: Leveraging Definitions and Semantic Expansion for Sexism Detection." pith.science (2026). https://pith.science/paper/WKSIEKUL

@misc{pith2026250606238,
  author       = {Pith},
  title        = {Pith review of: Explaining Matters: Leveraging Definitions and Semantic Expansion for Sexism Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WKSIEKUL}},
  note         = {Machine review of arXiv:2506.06238}
}

read the original abstract

The detection of sexism in online content remains an open problem, as harmful language disproportionately affects women and marginalized groups. While automated systems for sexism detection have been developed, they still face two key challenges: data sparsity and the nuanced nature of sexist language. Even in large, well-curated datasets like the Explainable Detection of Online Sexism (EDOS), severe class imbalance hinders model generalization. Additionally, the overlapping and ambiguous boundaries of fine-grained categories introduce substantial annotator disagreement, reflecting the difficulty of interpreting nuanced expressions of sexism. To address these challenges, we propose two prompt-based data augmentation techniques: Definition-based Data Augmentation (DDA), which leverages category-specific definitions to generate semantically-aligned synthetic examples, and Contextual Semantic Expansion (CSE), which targets systematic model errors by enriching examples with task-specific semantic features. To further improve reliability in fine-grained classification, we introduce an ensemble strategy that resolves prediction ties by aggregating complementary perspectives from multiple language models. Our experimental evaluation on the EDOS dataset demonstrates state-of-the-art performance across all tasks, with notable improvements of macro F1 by 1.5 points for binary classification (Task A) and 4.1 points for fine-grained classification (Task C).

Figures

Figures reproduced from arXiv: 2506.06238 by the authors.

**Figure 2.** Overview of the proposed pipeline. The system consists of four stages: pre-training, data augmentation [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Contextual Semantic Expansion (CSE) Exam [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Difference confusion matrix for Task C (with Definition-based Data Augmentation minus without Definition-based Data Augmentation) Positive values show where Definition-based Data Augmentation (DDA) increases counts relative to the baseline, negative values show reduc…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

53 extracted references · 34 canonical work pages

[1]

Gavin Abercrombie, Aiqi Jiang, Poppy Gerrard-abbott, Ioannis Konstas, and Verena Rieser. 2023. https://doi.org/10.18653/v1/2023.woah-1.17 Resources for automated identification of online gender-based violence: A systematic review . In The 7th Workshop on Online Abuse and Harms (WOAH), pages 170--186, Toronto, Canada. Association for Computational Linguistics

doi:10.18653/v1/2023.woah-1.17 2023
[2]

Gavin Abercrombie, Nikolas Vitsakis, Aiqi Jiang, and Ioannis Konstas. 2024. https://aclanthology.org/2024.nlperspectives-1.3 Revisiting annotation of online gender-based violence . In Proceedings of the 3rd Workshop on Perspectivist Approaches to NLP (NLPerspectives) @ LREC-COLING 2024, pages 31--41, Torino, Italia. ELRA and ICCL

work page 2024
[3]

Dina Almanea and Massimo Poesio. 2022. https://aclanthology.org/2022.lrec-1.244 A r MIS - the A rabic misogyny and sexism corpus with annotator subjective disagreements . In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 2282--2291, Marseille, France. European Language Resources Association

work page 2022
[4]

Helena Bonaldi, Yi Ling Chung, Gavin Abercrombie, and Marco Guerini. 2024. Nlp for counterspeech against hate: A survey and how-to guide. In Findings of the Association for Computational Linguistics, pages 3480--3499. Association for Computational Linguistics. Publisher Copyright: 2024 Association for Computational Linguistics.; 2024 Findings of the Assoc...

work page 2024
[5]

L. Breiman. 1996. https://api.semanticscholar.org/CorpusID:47328136 Bagging predictors . Machine Learning, 24:123--140

work page 1996
[6]

Tommaso Caselli, Valerio Basile, Jelena Mitrovi \'c , and Michael Granitzer. 2021. H ate BERT : Retraining BERT for abusive language detection in E nglish. In Proceedings of the 5th Workshop on Online Abuse and Harms (WOAH 2021), pages 17--25, Online. Association for Computational Linguistics

work page 2021
[7]

Tommaso Caselli, Valerio Basile, Jelena Mitrovi \'c , Inga Kartoziya, and Michael Granitzer. 2020. https://aclanthology.org/2020.lrec-1.760 I feel offended, don ' t be abusive! implicit/explicit messages in offensive and abusive language . In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 6193--6202, Marseille, France. Euro...

work page 2020
[8]

Le, and Christopher D

Kevin Clark, Minh-Thang Luong, Quoc V. Le, and Christopher D. Manning. 2020. ELECTRA : Pre-training text encoders as discriminators rather than generators. In Proceedings of the 8th International Conference on Learning Representations (ICLR 2020), Addis Ababa, Ethiopia

work page 2020

Show all 53 references

[9]

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. https://www.aclweb.org/anthology/N19-1423/ Bert: Pre-training of deep bidirectional transformers for language understanding . In Proceedings of the 2019 Conference of the North American Chapter of the Asso...

2019
[10]

Bosheng Ding, Chengwei Qin, Ruochen Zhao, Tianze Luo, Xinze Li, Guizhen Chen, Wenhan Xia, Junjie Hu, Anh Tuan Luu, and Shafiq Joty. 2024. https://arxiv.org/abs/2403.02990 Data augmentation using large language models: Data perspectives, learning paradigms and challenges . Prep...

2024 arXiv
[11]

Aditi Dutta, Susan Banducci, and Chico Q. Camargo. 2024. https://arxiv.org/abs/2409.20204 Divided by discipline? a systematic literature review on the quantification of online sexism and misogyny using a semi-automated approach . Preprint, arXiv:2409.20204

2024 arXiv
[12]

Weston Feely, Prabhakar Gupta, Manas Ranjan Mohanty, Timothy Chon, Tuhin Kundu, Vijit Singh, Sandeep Atluri, Tanya Roosta, Viviane Ghaderi, and Peter Schulam. 2023 a . https://doi.org/10.18653/v1/2023.semeval-1.175 QC on at S em E val-2023 task 10: Data augmentation and model ...

2023 doi
[13]

Weston Feely, Prabhakar Gupta, Manas Ranjan Mohanty, Timothy Chon, Tuhin Kundu, Vijit Singh, Sandeep Atluri, Tanya Roosta, Viviane Ghaderi, and Peter Schulam. 2023 b . QCon at S em E val-2023 task 10: Data augmentation and model ensembling for detection of online sexism. In Pr...

2023
[14]

Fraser, and Svetlana Kiritchenko

Rongchen Guo, Isar Nejadgholi, Hillary Dawkins, Kathleen C. Fraser, and Svetlana Kiritchenko. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.1090 Adaptable moral stances of large language models on sexist content: Implications for society and gender discourse . In Proceedin...

2024 doi
[15]

Pengcheng He, Xiaodong Liu, Jianfeng Gao, and Weizhu Chen. 2021. https://arxiv.org/abs/2006.03654 Deberta: Decoding-enhanced bert with disentangled attention . Preprint, arXiv:2006.03654

2021 arXiv
[16]

Md Saroar Jahan and Mourad Oussalah. 2023. https://doi.org/10.1016/j.neucom.2023.126232 A systematic review of hate speech automatic detection using natural language processing . Neurocomputing, 546:126232

2023
[17]

Aiqi Jiang, Nikolas Vitsakis, Tanvi Dinkar, Gavin Abercrombie, and Ioannis Konstas. 2024. https://arxiv.org/abs/2410.03543 Re-examining sexism and misogyny classification with annotator attitudes . Preprint, arXiv:2410.03543

2024 arXiv
[18]

Albert Q. Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lucile Saulnier, Lélio Renard Lavaud, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas ...

2023 arXiv
[19]

Sahrish Khan, Gabriele Pergola, and Arshad Jhumka. 2024. https://ceur-ws.org/Vol-3740/paper-99.pdf Multilingual sexism identification via fusion of large language models . In Joint Proceedings of the ACM IUI 2024 Workshops, volume 3740 of CEUR Workshop Proceedings, Grenoble, F...

2024
[20]

Hannah Kirk, Wenjie Yin, Bertie Vidgen, and Paul R \"o ttger. 2023. https://doi.org/10.18653/v1/2023.semeval-1.305 S em E val-2023 task 10: Explainable detection of online sexism

2023 doi
[21]

Sosuke Kobayashi. 2018. https://doi.org/10.18653/v1/N18-2072 Contextual augmentation: Data augmentation by words with paradigmatic relations . In Proceedings of the 2018 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language ...

2018 doi
[22]

Dong-Ho Lee, Hyundong Cho, Woojeong Jin, Jihyung Moon, Sungjoon Park, Paul R \"o ttger, Jay Pujara, and Roy Ka-wei Lee. 2024. https://doi.org/10.18653/v1/2024.woah-1.21 Improving covert toxicity detection by retrieving and generating references . In Proceedings of the 8th Work...

2024 doi
[23]

Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. https://arxiv.org/abs/1907.11692 Roberta: A robustly optimized bert pretraining approach . Preprint, arXiv:1907.11692

2019 arXiv
[24]

Junru Lu, Xingwei Tan, Gabriele Pergola, Lin Gui, and Yulan He. 2022. https://aclanthology.org/2022.findings-emnlp.176 Event-centric question answering via contrastive learning and invertible event transformation . In Findings of the Association for Computational Linguistics: ...

2022
[25]

Chen Lyu, , and Gabriele Pergola. 2024. https://aclanthology.org/2024.tsar-1.10/ S ci G is P y: a novel metric for biomedical text simplification via gist inference score . In Proceedings of the Third Workshop on Text Simplification, Accessibility and Readability (TSAR 2024), ...

2024
[26]

Edward Ma. 2019. Nlp augmentation. https://github.com/makcedward/nlpaug

2019
[27]

Subhabrata Mukherjee and Ahmed Hassan Awadallah. 2020. https://doi.org/10.18653/v1/2020.acl-main.202 X treme D istil: Multi-stage distillation for massive multilingual models . In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 22...

2020 doi
[28]

Roberto Navigli, Simone Conia, and Bj\" o rn Ross. 2023. https://doi.org/10.1145/3597307 Biases in large language models: Origins, inventory, and discussion . J. Data and Information Quality, 15(2)

2023 doi
[29]

Huy Nghiem and Hal Daum \'e Iii. 2024. https://doi.org/10.18653/v1/2024.findings-emnlp.343 H ate COT : An explanation-enhanced dataset for generalizable offensive speech detection via large language models . In Findings of the Association for Computational Linguistics: EMNLP 2...

2024 doi
[30]

Gabriele Pergola, Lin Gui, and Yulan He. 2019. Tdam: A T opic- D ependent A ttention M odel for S entiment A nalysis. Information Processing & Management, 56(6):102084

2019
[31]

Gabriele Pergola, Lin Gui, and Yulan He. 2021. https://doi.org/10.18653/v1/2021.naacl-main.228 A disentangled adversarial neural topic model for separating opinions from plots in user reviews . In Proceedings of the 2021 Conference of the North American Chapter of the Associat...

2021 doi
[32]

Sumegh Roychowdhury and Vikram Gupta. 2023. https://aclanthology.org/2023.findings-eacl.9 Data-efficient methods for improving hate speech detection . In Findings of the Association for Computational Linguistics: EACL 2023, pages 125--132, Dubrovnik, Croatia. Association for C...

2023
[33]

Isabel Segura-Bedmar. 2023 a . https://doi.org/10.18653/v1/2023.semeval-1.26 HULAT at S em E val-2023 task 10: Data augmentation for pre-trained transformers applied to the detection of sexism in social media . In Proceedings of the 17th International Workshop on Semantic Eval...

2023 doi
[34]

Isabel Segura-Bedmar. 2023 b . HULAT at S em E val-2023 task 10: Data augmentation for pre-trained transformers applied to the detection of sexism in social media. In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), pages 184--192, Toronto,...

2023
[35]

Enock, Tvesha Sippy, Jonathan Bright, Miranda Cross, Pica Johansson, Judy Wajcman, and Helen Z

Francesca Stevens, Florence E. Enock, Tvesha Sippy, Jonathan Bright, Miranda Cross, Pica Johansson, Judy Wajcman, and Helen Z. Margetts. 2024. https://arxiv.org/abs/2403.19037 Women are less comfortable expressing opinions online than men and report heightened fears for safety...

2024
[36]

Fahim Sufi. 2024. https://doi.org/10.3390/info15020099 Generative pre-trained transformer (gpt) in research: A systematic review on data augmentation . Information, 15(2)

2024 doi
[37]

Zhaoyue Sun, Gabriele Pergola, Byron Wallace, and Yulan He. 2024. Leveraging C hat GPT in pharmacovigilance event extraction: An empirical study. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 2: Short Pap...

2024
[38]

Ilya Sutskever, Oriol Vinyals, and Quoc V. Le. 2016. Sequence to sequence learning with neural networks. In Advances in Neural Information Processing Systems (NeurIPS), pages 3104--3112

2016
[39]

Xingwei Tan, Gabriele Pergola, and Yulan He. 2023. https://aclanthology.org/2023.eacl-main.80 Event temporal relation extraction with B ayesian translational model . In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics,...

2023
[40]

Xingwei Tan, Yuxiang Zhou, Gabriele Pergola, and Yulan He. 2024. Set-aligning framework for auto-regressive event temporal graph generation. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Tec...

2024
[41]

Xingwei Tan, Yuxiang Zhou, Gabriele Pergola, and Yulan He. 2025. https://aclanthology.org/2025.naacl-long.112/ Cascading large language models for salient event graph generation . In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association f...

2025
[42]

Nikolas Vitsakis, Amit Parekh, and Ioannis Konstas. 2024. https://arxiv.org/abs/2407.14259 Voices in a crowd: Searching for clusters of unique perspectives . Preprint, arXiv:2407.14259

2024 arXiv
[43]

Jason Wei and Kai Zou. 2019. https://doi.org/10.18653/v1/D19-1670 EDA : Easy data augmentation techniques for boosting performance on text classification tasks . In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International...

2019 doi
[44]

David H. Wolpert. 1992. Stacked generalization. Neural Networks, 5(2):241--259

1992
[45]

Albert Xu, Eshaan Pathak, Eric Wallace, Suchin Gururangan, Maarten Sap, and Dan Klein. 2021. https://doi.org/10.18653/v1/2021.naacl-main.190 Detoxifying language models risks marginalizing minority voices . In Proceedings of the 2021 Conference of the North American Chapter of...

2021 doi
[46]

Zhangyue Yin, Qiushi Sun, Cheng Chang, Qipeng Guo, Junqi Dai, Xuanjing Huang, and Xipeng Qiu. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.936 Exchange-of-thought: Enhancing large language model capabilities through cross-model communication . In Proceedings of the 2023 C...

2023 doi
[47]

Xinyang Zhang, Yury Malkov, Omar Florez, Serim Park, Brian McWilliams, Jiawei Han, and Ahmed El-Kishky. 2023. Twhin-bert: A socially-enriched pre-trained language model for multilingual tweet representations at twitter. In Proceedings of the 29th ACM SIGKDD Conference on Knowl...

2023
[48]

Ruochen Zhao, Xingxuan Li, Shafiq Joty, Chengwei Qin, and Lidong Bing. 2023. https://doi.org/10.18653/v1/2023.acl-long.320 Verify-and-edit: A knowledge-enhanced chain-of-thought framework . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguist...

2023 doi
[49]

Tianyun Zhong, Runhui Song, Xunyuan Liu, Juelin Wang, Boya Wang, and Binyang Li. 2023. https://doi.org/10.18653/v1/2023.semeval-1.287 UIRISC at S em E val-2023 task 10: Explainable detection of online sexism by ensembling fine-tuning language models . In Proceedings of the 17t...

2023 doi
[50]

Mengyuan Zhou. 2023. https://doi.org/10.18653/v1/2023.semeval-1.304 P ing A n L ife I nsurance at S em E val-2023 task 10: Using multi-task learning to better detect online sexism . In Proceedings of the 17th International Workshop on Semantic Evaluation (SemEval-2023), pages ...

2023 doi
[51]

Lixing Zhu, Gabriele Pergola, Lin Gui, Deyu Zhou, and Yulan He. 2021. https://doi.org/10.18653/v1/2021.acl-long.125 Topic-driven and knowledge-aware transformer for dialogue emotion detection . In Proceedings of the 59th Annual Meeting of the Association for Computational Ling...

2021 doi
[52]

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...
[53]

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 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Gavin Abercrombie, Aiqi Jiang, Poppy Gerrard-abbott, Ioannis Konstas, and Verena Rieser. 2023. https://doi.org/10.18653/v1/2023.woah-1.17 Resources for automated identification of online gender-based violence: A systematic review . In The 7th Workshop on Online Abuse and Harms (WOAH), pages 170--186, Toronto, Canada. Association for Computational Linguistics

doi:10.18653/v1/2023.woah-1.17 2023

[2] [2]

Gavin Abercrombie, Nikolas Vitsakis, Aiqi Jiang, and Ioannis Konstas. 2024. https://aclanthology.org/2024.nlperspectives-1.3 Revisiting annotation of online gender-based violence . In Proceedings of the 3rd Workshop on Perspectivist Approaches to NLP (NLPerspectives) @ LREC-COLING 2024, pages 31--41, Torino, Italia. ELRA and ICCL

work page 2024

[3] [3]

Dina Almanea and Massimo Poesio. 2022. https://aclanthology.org/2022.lrec-1.244 A r MIS - the A rabic misogyny and sexism corpus with annotator subjective disagreements . In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 2282--2291, Marseille, France. European Language Resources Association

work page 2022

[4] [4]

Helena Bonaldi, Yi Ling Chung, Gavin Abercrombie, and Marco Guerini. 2024. Nlp for counterspeech against hate: A survey and how-to guide. In Findings of the Association for Computational Linguistics, pages 3480--3499. Association for Computational Linguistics. Publisher Copyright: 2024 Association for Computational Linguistics.; 2024 Findings of the Assoc...

work page 2024

[5] [5]

L. Breiman. 1996. https://api.semanticscholar.org/CorpusID:47328136 Bagging predictors . Machine Learning, 24:123--140

work page 1996

[6] [6]

Tommaso Caselli, Valerio Basile, Jelena Mitrovi \'c , and Michael Granitzer. 2021. H ate BERT : Retraining BERT for abusive language detection in E nglish. In Proceedings of the 5th Workshop on Online Abuse and Harms (WOAH 2021), pages 17--25, Online. Association for Computational Linguistics

work page 2021

[7] [7]

Tommaso Caselli, Valerio Basile, Jelena Mitrovi \'c , Inga Kartoziya, and Michael Granitzer. 2020. https://aclanthology.org/2020.lrec-1.760 I feel offended, don ' t be abusive! implicit/explicit messages in offensive and abusive language . In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 6193--6202, Marseille, France. Euro...

work page 2020

[8] [8]

Le, and Christopher D

Kevin Clark, Minh-Thang Luong, Quoc V. Le, and Christopher D. Manning. 2020. ELECTRA : Pre-training text encoders as discriminators rather than generators. In Proceedings of the 8th International Conference on Learning Representations (ICLR 2020), Addis Ababa, Ethiopia

work page 2020