Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Conversations Gone Awry, But Then? Evaluating Conversational Forecasting Models

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A standardized benchmark for derailment forecasting finds decoder-based LLMs are the only models to pass 70% accuracy.

desk verdict A genuinely useful evaluation framework and expanded dataset, but the model-ranking and Recovery claims need error bars before they carry weight. read the letter →

arxiv 2507.19470 v1 pith:LOP5TFUO submitted 2025-07-25 cs.CL cs.HC

Son Quoc Tran , Tushaar Gangavarapu , Nicholas Chernogor , Jonathan P. Chang , Cristian Danescu-Niculescu-Mizil This is my paper
classification cs.CLcs.HC
keywords conversationalforecastingconversationderailmentpredictionevaluationbenchmarkforecastrecoverylargelanguagemodelsthresholdtuningonlinemoderationChangeMyView
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 a single, reliable way to evaluate conversational forecasting models—systems that predict, in real time, whether an online conversation will derail into a personal attack. It introduces the first uniform evaluation framework and a new large dataset, then uses the framework to run a head-to-head comparison of thirteen models, from small recurrent networks to billion-parameter generative LLMs. The central finding is that decoder-based generative LLMs are the only models that surpass 70% accuracy on the large benchmark, with the best model reaching 71.0% accuracy. The paper also argues that existing classification metrics miss a key behavior—whether a model can revise a wrong prediction as the conversation unfolds—and proposes a new metric to capture that. A sympathetic reader should care because reliable comparisons and a metric for forecast revision are prerequisites for building systems that actually help moderators and conversation participants intervene before things go wrong.

What carries the argument

The load-bearing mechanism is the Forecast Recovery metric, computed as Recovery = CR/N − IR/N, where CR is the number of correct recoveries (the model first triggers a derailment forecast, revises it to non-derailment, and the conversation indeed does not derail) and IR is the number of incorrect recoveries (the model revises, but the conversation does derail). The paper proves that this metric equals the difference between two accuracies: the accuracy of the conversation-level forecast evaluated at the moment the model triggers, minus the accuracy of a hypothetical classification formulation that only looks at the model's final prediction. The metric is designed to be agnostic to why a recovery happens—whether the model made an initial error or the conversation actually recovered—because both cases benefit a real-time assistant. The framework also standardizes training by using a snapshot of the conversation before the final comment and by tuning the decision threshold on the development split, which the paper shows can change accuracy by 1.3 to 3.6 points.

What would settle it

Re-run the benchmark on a version of CGA-CMV-large that includes conversations with deleted utterances, recovering the deleted text wherever possible from moderation logs or public archives; if the accuracy ranking changes (for instance, if Gemma2 9B falls below 70% or Mistral 7B's Recovery advantage disappears), then the deletion filter is not neutral and the reported results do not generalize to unfiltered real-world conversations.

Watch

Extended reading notes

Core claim

The paper claims that prior evaluations of conversation-derailment forecasting were not directly comparable because they used different training setups, different assumptions about conversation length, and sometimes inconsistent threshold practices. To fix this, the authors build a modular evaluation framework, extend the ChangeMyView derailment dataset to 19,578 conversations collected through 2022, and benchmark thirteen models under identical conditions. On the large benchmark, decoder-based generative LLMs are the only models to exceed 70% accuracy: Gemma2 9B reaches 71.0% accuracy and 72.3 F1, with Mistral 7B and Phi4 14B statistically close. The paper also introduces a new metric, Forecast Recovery, defined as the fraction of correct recoveries minus the fraction of incorrect recoveries, where a recovery is the model first predicting derailment and later changing its forecast to non-derailment. This metric exposes differences that accuracy and F1 hide: for example, on the large benchmark Mistral 7B outperforms Gemma2 9B on Recovery even though their accuracy and F1 are similar, suggesting the smaller model is better at recognizing when a tense conversation is de-escalating.

Load-bearing premise

The benchmark's validity rests on the assumption that excluding every conversation that contains a deleted message—either deleted by users or by Reddit's automated systems—does not systematically bias the data, and the paper does not report how many conversations were removed or whether the removal correlates with derailment.

Editorial extensions

If this is right

  • Direct comparisons between architectures for derailment forecasting are now possible under identical conditions, making reported results from different papers more interpretable.
  • Decoder-based generative LLMs appear to be the strongest current approach for this task, at least on data from ChangeMyView conversations collected through 2022.
  • Model rankings change when Forecast Recovery is considered: Mistral 7B overtakes the higher-accuracy Gemma2 9B, which matters for choosing a model for real-time intervention tools.
  • Threshold tuning is unusually important for this forecasting task; failing to tune the threshold on a development split can lower accuracy by several points and likely explains some discrepancies in prior work.
  • The framework is designed to be extendable to other forecasting tasks, such as predicting prosocial outcomes, so the same evaluation machinery can be reused beyond derailment.

Reading between the lines

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

  • Forecast Recovery could be applied to any online forecasting task where a model must continuously update its belief, such as predicting persuasion success, controversy, or decision outcomes in team discussions.
  • The metric might be gameable: a model that deliberately triggers early on any sign of tension and then revises could inflate its Recovery score without being genuinely better at reading conversation dynamics, so future work may need to pair Recovery with costs or calibration.
  • The deletion-filtering choice, which removes conversations containing deleted utterances, could interact with the Recovery metric if deleted messages are disproportionately hostile, so the reported Recovery values may not generalize to unfiltered moderation settings.
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

3 major / 5 minor

Summary. The paper proposes a standardized evaluation framework for conversational forecasting, introduces an expanded CGA-CMV-large dataset (19,578 conversations from 2015-2022), benchmarks 13 models ranging from CRAFT to 14B-parameter decoder LLMs, and introduces a Forecast Recovery metric intended to capture whether a model revises an incorrect derailment prediction as the conversation evolves. On CGA-CMV-large, the paper reports that decoder-based generative LLMs are the only models surpassing 70% accuracy and Gemma2 9B is the best on accuracy (71.0%), with Mistral 7B best on Recovery (+2.9). The framework is released as part of ConvoKit, and the paper includes an ablation showing that Recovery distinguishes context-stripped from full-context variants more clearly than accuracy or F1.

Significance. If the claims hold, the framework would be a useful common testbed for conversational forecasting, and the Recovery metric provides a clean way to decompose the gap between conversation-level and final-utterance predictions. The mathematical identity in Appendix C.2 (Recovery equals the difference between the classification-formulation accuracy and the forecasting-formulation accuracy) is correct and clearly presented, and the authors deserve credit for releasing the framework, reporting hyperparameters and licenses, and systematically reproducing prior baselines. The main value is the breadth of the survey and the standardization of evaluation, rather than a fundamentally new modeling idea. The significance is solid but contingent on the reliability of the reported comparisons, which currently lack variance estimates, and on the representativeness of the filtered dataset.

major comments (3)
  1. [Section 4, Table 1] All headline results are point estimates averaged over five random seeds, with no standard deviations, confidence intervals, or per-seed values reported. The central empirical claims—that decoder-based generative LLMs are the only models surpassing 70% accuracy and that Gemma2 9B is SOTA (71.0% vs 70.7% Mistral and 70.5% Phi4; Recovery +2.9 Mistral vs +1.8 Gemma2)—are therefore not statistically supported. Because the framework's stated purpose is 'direct and reliable comparisons,' the absence of uncertainty quantification is load-bearing; please report per-seed results and variance-aware comparisons, and qualify the Recovery-based recommendation in Section 4 that Mistral 7B may be better suited for real-time interactive applications.
  2. [Section 3, footnote 3; Appendix A] The benchmark excludes every conversation containing deleted utterances (user-deleted or Reddit-removed for reasons other than Rule 2), described as a 'relatively minor fix,' but no count, fraction, or label distribution of the excluded conversations is given. Deleted messages are often removed precisely because they are hostile or rule-violating, so this filtering may systematically underrepresent derailment and change task difficulty. Please quantify the number and fraction of excluded conversations, report the label balance before and after filtering, and provide a robustness analysis or a statement of why the filtering cannot bias the comparisons.
  3. [Appendix C.2] The identity Recovery = (CR − IR)/N = Acc(final-utterance) − Acc(conversation-level) is derived cleanly and is correct, but it shows that the new metric is an exact linear combination of two existing accuracy quantities. The paper should state this equivalence explicitly and discuss what incremental information Recovery provides beyond reporting both accuracies; as written, the claim that traditional metrics cannot capture recovery dynamics overstates the metric's independence, even though the CR/IR decomposition is useful for interpretation.
minor comments (5)
  1. [Appendix C.1] There is a typo in the phrase 'thenrecovers' in the discussion of forecast recovery; it should read 'then recovers.'
  2. [Section 3] The mean horizon metric (Mean H) is used throughout Tables 1-3 but is never formally defined in the text; a one-sentence definition or explicit pointer to the original definition would improve readability.
  3. [Table 7] Training times are reported only for CGA-CMV-large; reporting approximate training times for CGA-Wiki and CGA-CMV-legacy would aid reproducibility.
  4. [Section 3] The paper should state clearly whether the CGA-CMV-large dataset will be released publicly alongside the ConvoKit framework, since the benchmark's usefulness depends on other researchers being able to access the data.
  5. [Tables 1-3] The bold formatting is inconsistent across tables: Table 1 has no bold entries, while Tables 2 and 3 bold per-group or global best values; making the formatting uniform and indicating the global best in each table would help readers.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the benchmark results are measured on held-out data, and the Recovery metric is an independently defined score whose accuracy-difference identity is a transparent derivation, not an input-dependent prediction.

full rationale

The paper's central claims—a standardized CGA evaluation framework, the SOTA status of decoder-based generative LLMs, and the utility of the Forecast Recovery metric—are supported by experiments on fixed train/validation/test splits with independently reproduced baselines. The Recovery metric is defined directly from counts of correct and incorrect recoveries (Recovery = CR/N - IR/N), and Appendix C.2 then transparently derives the identity Recovery = Acc_forecast - Acc_classification. This is a post-hoc mathematical equivalence, not a definition of the metric in terms of the result it is used to support, so it does not make the evaluation circular. The CGA-CMV-large dataset extends the authors' own prior CGA-CMV dataset and is released through their own ConvoKit toolkit, and the training approach follows prior work by Chang and Danescu-Niculescu-Mizil (2019b); these are self-citations, but the benchmark results are measured on held-out data and compared against independent model implementations, so no claim reduces to its own inputs. Threshold tuning is performed on the development split only, with an ablation quantifying its effect, so no fitted parameter is renamed as a prediction. The paper's acknowledged limitations—coarse recovery granularity, text-only evaluation, and the deleted-utterance filtering in Section 3—are data-validity and scope concerns, not circularity. Overall, the derivation chain is self-contained with respect to the paper's empirical claims.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

The central claims rest on data-label validity and the transfer of snapshot-trained models to online forecasting. No new theoretical entities are introduced; the only fitted quantity is the per-model triggering threshold.

free parameters (1)
  • Triggering threshold T = Per-model; 28 of 30 optimal thresholds exceed 0.5, mean 0.65, std 0.08 (Appendix D)
    The threshold converts per-utterance derailment probabilities into a conversation-level trigger decision. It is tuned on the development split to maximize accuracy for each model, and it directly affects every reported metric including Forecast Recovery.
assumptions (4)
  • domain assumption Moderator-removal labels in CGA-CMV reliably indicate conversation-level derailment (personal attack).
    Section 3 states the labels are based on comments removed for violating Rule 2. Earlier turns may already contain derailment, acknowledged as a source of noise.
  • domain assumption Excluding conversations with deleted utterances does not materially bias the benchmark.
    Section 3 footnote 3 and Appendix A remove these conversations but do not quantify the fraction or the resulting distribution shift.
  • domain assumption Fine-tuning on a single snapshot before the final utterance transfers to online forecasting at every timestamp.
    Section 4 and Figure 1 describe training on the final snapshot; the training-inference mismatch is discussed in Appendix D but the transfer is assumed.
  • domain assumption The no-context variant isolates the effect of conversational context.
    Appendix C.3 feeds only the most recent utterance to a model trained on full-context snapshots, so the observed difference could be due to input-format mismatch.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Conversations Gone Awry, But Then? Evaluating Conversational Forecasting Models." pith.science (2026). https://pith.science/paper/LOP5TFUO

@misc{pith2026250719470,
  author       = {Pith},
  title        = {Pith review of: Conversations Gone Awry, But Then? Evaluating Conversational Forecasting Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LOP5TFUO}},
  note         = {Machine review of arXiv:2507.19470}
}
read the original abstract

We often rely on our intuition to anticipate the direction of a conversation. Endowing automated systems with similar foresight can enable them to assist human-human interactions. Recent work on developing models with this predictive capacity has focused on the Conversations Gone Awry (CGA) task: forecasting whether an ongoing conversation will derail. In this work, we revisit this task and introduce the first uniform evaluation framework, creating a benchmark that enables direct and reliable comparisons between different architectures. This allows us to present an up-to-date overview of the current progress in CGA models, in light of recent advancements in language modeling. Our framework also introduces a novel metric that captures a model's ability to revise its forecast as the conversation progresses.

Figures

Figures reproduced from arXiv: 2507.19470 by the authors.

Figure 1
Figure 1. Illustration of Training (left) and Testing (right) CGA models. During evaluation, models must generate forecasts for all timestamps in a conversation. However, models are trained only to distinguish between derailing and non-derailing conversations at the last timestamp (t = n − 1) of each conversation. As described in Section 4, the most effective training strategy is to use snapshots of the conver￾sation taken at… view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Wait! There's a Way Out: A Decision Mechanism for Forecasting Conversational Derailment

    cs.CL 2026-05 unverdicted novelty 7.0 of 10

    A deferral mechanism using forward-looking simulations reduces false positives in derailment forecasting by selectively waiting when recovery paths appear plausible.

Reference graph

Works this paper leans on

42 extracted references · 14 canonical work pages · cited by 1 Pith paper

  1. [1]

    Hewett, Mojan Javaheripi, Piero Kauffmann, James R

    Marah Abdin, Jyoti Aneja, Harkirat Behl, Sébastien Bubeck, Ronen Eldan, Suriya Gunasekar, Michael Harrison, Russell J. Hewett, Mojan Javaheripi, Piero Kauffmann, James R. Lee, Yin Tat Lee, Yuanzhi Li, Weishung Liu, Caio C. T. Mendes, Anh Nguyen, Eric Price, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Xin Wang, Rachel Ward, Yue Wu, Dingli Yu,...

    arXiv 2024

  2. [3]

    Enas Altarawneh, Ameeta Agrawal, Michael Jenkin, and Manos Papagelis. 2023 b . https://doi.org/10.18653/v1/2023.woah-1.16 Conversation Derailment Forecasting with Graph Convolutional Networks . In The 7th Workshop on Online Abuse and Harms ( WOAH ) , pages 160--169, Toronto, Canada. Association for Computational Linguistics

    doi:10.18653/v1/2023.woah-1.16 2023

  3. [4]

    Tim Althoff, Kevin Clark, and Jure Leskovec. 2016. https://doi.org/10.1162/tacl_a_00111 Large-scale analysis of counseling conversations: An application of natural language processing to mental health . Transactions of the Association for Computational Linguistics, 4:463--476

    doi:10.1162/tacl_a_00111 2016

  4. [5]

    Jiajun Bao, Junjie Wu, Yiming Zhang, Eshwar Chandrasekharan, and David Jurgens. 2021. https://doi.org/10.1145/3442381.3450122 Conversations gone alright: Quantifying and predicting prosocial outcomes in online conversations . In Proceedings of the Web Conference 2021, WWW '21, page 1134–1145, New York, NY, USA. Association for Computing Machinery

    work page arXiv 2021

  5. [6]

    Jonathan Chang and Cristian Danescu-Niculescu-Mizil. 2019 a . https://doi.org/10.1145/3308558.3313638 Trajectories of blocked community members: Redemption, recidivism and departure . In The World Wide Web Conference, WWW '19, page 184–195, New York, NY, USA. Association for Computing Machinery

    arXiv 2019

  6. [7]

    Chang, Caleb Chiam, Liye Fu, Andrew Wang, Justine Zhang, and Cristian Danescu-Niculescu-Mizil

    Jonathan P. Chang, Caleb Chiam, Liye Fu, Andrew Wang, Justine Zhang, and Cristian Danescu-Niculescu-Mizil. 2020. https://doi.org/10.18653/v1/2020.sigdial-1.8 C onvo K it: A toolkit for the analysis of conversations . In Proceedings of the 21th Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 57--60, 1st virtual meeting. Associ...

    doi:10.18653/v1/2020.sigdial-1.8 2020

  7. [8]

    Chang and Cristian Danescu-Niculescu-Mizil

    Jonathan P. Chang and Cristian Danescu-Niculescu-Mizil. 2019 b . https://doi.org/10.18653/v1/D19-1481 Trouble on the horizon: Forecasting the derailment of online conversations as they develop . In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processin...

    doi:10.18653/v1/d19-1481 2019

  8. [9]

    Chang, Charlotte Schluger, and Cristian Danescu-Niculescu-Mizil

    Jonathan P. Chang, Charlotte Schluger, and Cristian Danescu-Niculescu-Mizil. 2022. https://doi.org/10.1145/3555603 Thread with caution: Proactively helping users assess and deescalate tension in their online discussions . Proc. ACM Hum.-Comput. Interact., 6(CSCW2)

    work page doi:10.1145/3555603 2022

Show all 42 references
  1. [10]

    Curhan and Alex Pentland

    Jared R. Curhan and Alex Pentland. 2007. https://doi.org/10.1037/0021-9010.92.3.802 Thin slices of negotiation: Predicting outcomes from conversational dynamics within the first 5 minutes. Journal of Applied Psychology, 92:802--811

    2007 doi

  2. [11]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. https://doi.org/10.18653/v1/N19-1423 BERT : Pre-training of deep bidirectional transformers for language understanding . In Proceedings of the 2019 Conference of the North A merican Chapter of the Associat...

    2019 doi

  3. [12]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Art...

    2024 arXiv

  4. [13]

    Pengcheng He, Jianfeng Gao, and Weizhu Chen. 2023. https://openreview.net/forum?id=sE7-XhLxHA De BERT av3: Improving de BERT a using ELECTRA -style pre-training with gradient-disentangled embedding sharing . In The Eleventh International Conference on Learning Representations

    2023

  5. [14]

    Jack Hessel and Lillian Lee. 2019. https://doi.org/10.18653/v1/N19-1166 Something ' s brewing! early prediction of controversy-causing posts from discussion features . In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Lin...

    2019 doi

  6. [15]

    Smith, and Mari Ostendorf

    Yushi Hu, Chia-Hsuan Lee, Tianbao Xie, Tao Yu, Noah A. Smith, and Mari Ostendorf. 2022. https://doi.org/10.18653/v1/2022.findings-emnlp.193 In-context learning for few-shot dialogue state tracking . In Findings of the Association for Computational Linguistics: EMNLP 2022, page...

    2022 doi

  7. [16]

    Yiqing Hua, Cristian Danescu-Niculescu-Mizil, Dario Taraborelli, Nithum Thain, Jeffery Sorensen, and Lucas Dixon. 2018. https://doi.org/10.18653/v1/D18-1305 W iki C onv: A corpus of the complete conversational history of a large online collaborative community . In Proceedings ...

    2018 doi

  8. [17]

    Piotr Janiszewski, Mateusz Lango, and Jerzy Stefanowski. 2021. Time aspect in making an actionable prediction of a conversation breakdown. In Machine Learning and Knowledge Discovery in Databases. Applied Data Science Track, pages 351--364, Cham. Springer International Publishing

    2021

  9. [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

  10. [19]

    Weld, Luke Zettlemoyer, and Omer Levy

    Mandar Joshi, Danqi Chen, Yinhan Liu, Daniel S. Weld, Luke Zettlemoyer, and Omer Levy. 2020. https://doi.org/10.1162/tacl_a_00300 S pan BERT : Improving pre-training by representing and predicting spans . Transactions of the Association for Computational Linguistics, 8:64--77

    2020 doi

  11. [20]

    David Jurgens, Libby Hemphill, and Eshwar Chandrasekharan. 2019. https://doi.org/10.18653/v1/P19-1357 A just and comprehensive strategy for using NLP to address online abuse . In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 365...

    2019 doi

  12. [21]

    Yova Kementchedjhieva and Anders S gaard. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.624 Dynamic forecasting of conversation derailment . In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 7915--7919, Online and Punta Cana, ...

    2021 doi

  13. [22]

    Svetlana Kiritchenko, Isar Nejadgholi, and Kathleen C. Fraser. 2021. https://doi.org/10.1613/jair.1.12590 Confronting Abusive Language Online : A Survey from the Ethical and Human Rights Perspective . Journal of Artificial Intelligence Research, 71:431--478

    2021 doi

  14. [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

  15. [24]

    Elijah Mayfield and Alan W. Black. 2019. https://doi.org/10.1145/3359308 Analyzing wikipedia deletion debates with a group decision-making forecast model . Proc. ACM Hum.-Comput. Interact., 3(CSCW)

    2019 doi

  16. [25]

    Vlad Niculae and Cristian Danescu-Niculescu-Mizil. 2016. https://doi.org/10.18653/v1/N16-1070 Conversational markers of constructive discussions . In Proceedings of the 2016 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Langu...

    2016 doi

  17. [26]

    Ji Ho Park, Jamin Shin, and Pascale Fung. 2018. https://doi.org/10.18653/v1/D18-1302 Reducing Gender Bias in Abusive Language Detection . In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing , pages 2799--2804, Brussels, Belgium. Associatio...

    2018 doi

  18. [27]

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas K\" o pf, Edward Yang, Zach DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu...

    2019

  19. [28]

    Morgane Riviere, Shreya Pathak, Pier Giuseppe Sessa, Cassidy Hardin, Surya Bhupatiraju, Léonard Hussenot, Thomas Mesnard, Bobak Shahriari, Alexandre Ramé, Johan Ferret, Peter Liu, Pouya Tafti, Abe Friesen, Michelle Casbon, Sabela Ramos, Ravin Kumar, Charline Le Lan, Sammy Jero...

    2024 arXiv

  20. [29]

    Maarten Sap, Dallas Card, Saadia Gabriel, Yejin Choi, and Noah A. Smith. 2019. https://doi.org/10.18653/v1/P19-1163 The Risk of Racial Bias in Hate Speech Detection . In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics , pages 1668--1678,...

    2019 doi

  21. [30]

    Chang, Cristian Danescu-Niculescu-Mizil, and Karen Levy

    Charlotte Schluger, Jonathan P. Chang, Cristian Danescu-Niculescu-Mizil, and Karen Levy. 2022. https://doi.org/10.1145/3555095 Proactive moderation of online discussions: Existing practices and the potential for algorithmic support . Proc. ACM Hum.-Comput. Interact., 6(CSCW2)

    2022 doi

  22. [31]

    Ana Smith. 2023. https://www.proquest.com/docview/2827129304/abstract/94C45D7E632D41AFPQ/1 Leveraging Context Documents for Social Natural Language Processing . phdthesis, Cornell University. ISBN : 9798379712426

    2023

  23. [32]

    Chenhao Tan, Vlad Niculae, Cristian Danescu-Niculescu-Mizil, and Lillian Lee. 2016. https://doi.org/10.1145/2872427.2883081 Winning arguments: Interaction dynamics and persuasion strategies in good-faith online discussions . In Proceedings of the 25th International Conference ...

    2016

  24. [33]

    Henning Wachsmuth, Shahbaz Syed, and Benno Stein. 2018. https://doi.org/10.18653/v1/P18-1023 Retrieval of the best counterargument without prior topic knowledge . In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)...

    2018 doi

  25. [34]

    Tony Wang, Amy S Bruckman, and Diyi Yang. 2025. https://doi.org/10.1145/3711089 The Practice of Online Peer Counseling and the Potential for AI - Powered Support Tools . Proc. ACM Hum.-Comput. Interact., 9(CSCW)

    2025 doi

  26. [35]

    Michael Wiegand, Josef Ruppenhofer, and Thomas Kleinbauer. 2019. https://doi.org/10.18653/v1/N19-1060 Detection of Abusive Language : the Problem of Biased Datasets . In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Lingu...

    2019 doi

  27. [36]

    Robert Williamson and Aditya Menon. 2019. https://proceedings.mlr.press/v97/williamson19a.html Fairness risk measures . In Proceedings of the 36th International Conference on Machine Learning , pages 6786--6797. PMLR. ISSN: 2640-3498

    2019

  28. [37]

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mari...

    2020 doi

  29. [38]

    Diyi Yang, Jiaao Chen, Zichao Yang, Dan Jurafsky, and Eduard Hovy. 2019. https://doi.org/10.18653/v1/N19-1364 Let ' s make your request more persuasive: Modeling persuasive strategies via semi-supervised neural nets on crowdfunding platforms . In Proceedings of the 2019 Confer...

    2019 doi

  30. [39]

    Jiaqing Yuan and Munindar P. Singh. 2023. https://doi.org/10.1609/icwsm.v17i1.22200 Conversation modeling to predict derailment . Proceedings of the International AAAI Conference on Web and Social Media, 17(1):926--935

    2023 doi

  31. [40]

    Justine Zhang, Jonathan Chang, Cristian Danescu-Niculescu-Mizil, Lucas Dixon, Yiqing Hua, Dario Taraborelli, and Nithum Thain. 2018 a . https://doi.org/10.18653/v1/P18-1125 Conversations gone awry: Detecting early signs of conversational failure . In Proceedings of the 56th An...

    2018 doi

  32. [41]

    Justine Zhang, Cristian Danescu-Niculescu-Mizil, Christina Sauper, and Sean J. Taylor. 2018 b . https://doi.org/10.1145/3274467 Characterizing online public discussions through patterns of participant interactions . In Proceedings of CSCW

    2018 doi

  33. [42]

    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...

  34. [43]

    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 6, 2026 · model on record in the stance chip above.