EMNLP Findings2023
High-quality argumentative information in low resources approaches improve counter-narrative generation
Damián Ariel Furman, Pablo Torres, José A. Rodríguez, Diego Letzen, Maria Vanina Martinez, Laura Alonso Alemany
Abstract
It has been shown that high quality fine-tuning boosts the performance of language models, even if the size of the fine-tuning is small. In this work we show how highly targeted fine-tuning improves the task of hate speech counter-narrative generation in user-generated text, even for very small sizes of training (1722 counter-narratives for English and 355 for Spanish). Providing a small subset of examples focusing on single argumentative strategies, together with the argumentative analysis relevant to that strategy, yields counter-narratives that are as satisfactory as providing the whole set of counter-narratives. We also show that a good base model is required for the fine-tuning to have a positive impact. Indeed, for Spanish, the counter-narratives obtained without fine-tuning are mostly unacceptable, and, while fine-tuning improves their overall quality, the performance still remains quite unsatisfactory.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
4 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Beyond Static Benchmarks: Synthesizing Harmful Content via Persona-based Simulation for Robust Evaluation cites Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate Speech
- Beyond Static Benchmarks: Synthesizing Harmful Content via Persona-based Simulation for Robust Evaluation cites Counterspeeches up my sleeve! Intent Distribution Learning and Persistent Fusion for Intent-Conditioned Counterspeech Generation
- Fact-based Counter Narrative Generation to Combat Hate Speech cites Generating Counter Narratives against Online Hate Speech: Data and Strategies
- Fact-based Counter Narrative Generation to Combat Hate Speech cites Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate Speech
- Fact-based Counter Narrative Generation to Combat Hate Speech cites Human-Machine Collaboration Approaches to Build a Dialogue Dataset for Hate Speech Countering
- Fact-based Counter Narrative Generation to Combat Hate Speech cites Counterspeeches up my sleeve! Intent Distribution Learning and Persistent Fusion for Intent-Conditioned Counterspeech Generation
- Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate Speech cites Generating Counter Narratives against Online Hate Speech: Data and Strategies
- Human-Machine Collaboration Approaches to Build a Dialogue Dataset for Hate Speech Countering cites Generating Counter Narratives against Online Hate Speech: Data and Strategies
- Human-Machine Collaboration Approaches to Build a Dialogue Dataset for Hate Speech Countering cites Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate Speech
- Counterspeeches up my sleeve! Intent Distribution Learning and Persistent Fusion for Intent-Conditioned Counterspeech Generation cites Generating Counter Narratives against Online Hate Speech: Data and Strategies
- Counterspeeches up my sleeve! Intent Distribution Learning and Persistent Fusion for Intent-Conditioned Counterspeech Generation cites Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate Speech
Cited by top-tier papers4
- Argumentative Large Language Models for Explainable and Contestable Claim VerificationGabriel Freedman, Adam Dejl, Deniz Gorur, Xiang Yin et al.AAAI 2025 · 31 citations
- Fact-based Counter Narrative Generation to Combat Hate SpeechBrian Wilk, Homaira Huda Shomee, Suman Kalyan Maity, Sourav MedyaWWW 2025 · 6 citations
- Let's discuss! Quality Dimensions and Annotated Datasets for Computational Argument Quality AssessmentRositsa V. Ivanova, Thomas Huber, Christina NiklausEMNLP 2024 · 2 citations
- Beyond Static Benchmarks: Synthesizing Harmful Content via Persona-based Simulation for Robust EvaluationHuije Lee, Jisu Shin, Hoyun Song, Changgeon Ko et al.ACL 2026
Builds on4
- Human-Machine Collaboration Approaches to Build a Dialogue Dataset for Hate Speech CounteringHelena Bonaldi, Sara Dellantonio, Serra Sinem Tekiroglu, Marco GueriniEMNLP 2022 · 18 citations
- Generating Counter Narratives against Online Hate Speech: Data and StrategiesSerra Sinem Tekiroglu, Yi-Ling Chung, Marco GueriniACL 2020 · 13 citations
- Counterspeeches up my sleeve! Intent Distribution Learning and Persistent Fusion for Intent-Conditioned Counterspeech GenerationRishabh Gupta, Shaily Desai, Manvi Goel, Anil Bandhakavi et al.ACL 2023 · 10 citations
- Human-in-the-Loop for Data Collection: a Multi-Target Counter Narrative Dataset to Fight Online Hate SpeechMargherita Fanton, Helena Bonaldi, Serra Sinem Tekiroglu, Marco GueriniACL 2021
Related papers
- Using Pre-Trained Language Models for Producing Counter Narratives Against Hate Speech: a Comparative StudySerra Sinem Tekiroglu, Helena Bonaldi, Margherita Fanton, Marco GueriniACL Findings 2022 · 16 citations
- Is Safer Better? The Impact of Guardrails on the Argumentative Strength of LLMs in Hate Speech CounteringHelena Bonaldi, Greta Damo, Nicolás Benjamín Ocampo, Elena Cabrio et al.EMNLP 2024 · 2 citations
- Outcome-Constrained Large Language Models for Countering Hate SpeechLingzi Hong, Pengcheng Luo, Eduardo Blanco, Xiaoying SongEMNLP 2024 · 5 citations
- Towards Knowledge-Grounded Counter Narrative Generation for Hate SpeechYi-Ling Chung, Serra Sinem Tekiroglu, Marco GueriniACL Findings 2021 · 29 citations
- Fight Fire with Fire: Fine-tuning Hate Detectors using Large Samples of Generated Hate SpeechTomer Wullach, Amir Adler, Einat MinkovEMNLP Findings 2021 · 33 citations
