@inproceedings{wadhwa-etal-2025-northeastern,
    title = "Northeastern Uni at Multilingual Counterspeech Generation: Enhancing Counter Speech Generation with {LLM} Alignment through Direct Preference Optimization",
    author = "Wadhwa, Sahil  and
      Xu, Chengtian  and
      Chen, Haoming  and
      Mahalingam, Aakash  and
      Kar, Akankshya  and
      Chaudhary, Divya",
    editor = "Bonaldi, Helena  and
      Vallecillo-Rodr{\'i}guez, Mar{\'i}a Estrella  and
      Zubiaga, Irune  and
      Montejo-R{\'a}ez, Arturo  and
      Soroa, Aitor  and
      Mart{\'i}n-Valdivia, Mar{\'i}a Teresa  and
      Guerini, Marco  and
      Agerri, Rodrigo",
    booktitle = "Proceedings of the First Workshop on Multilingual Counterspeech Generation",
    month = jan,
    year = "2025",
    address = "Abu Dhabi, UAE",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/fix-sig-urls/2025.mcg-1.3/",
    pages = "19--28",
    abstract = "The automatic generation of counter-speech (CS) is a critical strategy for addressing hate speech by providing constructive and informed responses. However, existing methods often fail to generate high-quality, impactful, and scalable CS, particularly across diverse lin- guistic contexts. In this paper, we propose a novel methodology to enhance CS generation by aligning Large Language Models (LLMs) using Supervised Fine-Tuning (SFT) and Di- rect Preference Optimization (DPO). Our ap- proach leverages DPO to align LLM outputs with human preferences, ensuring contextu- ally appropriate and linguistically adaptable responses. Additionally, we incorporate knowl- edge grounding to enhance the factual accuracy and relevance of generated CS. Experimental results demonstrate that DPO-aligned models significantly outperform SFT baselines on CS benchmarks while scaling effectively to mul- tiple languages. These findings highlight the potential of preference-based alignment tech- niques to advance CS generation across var- ied linguistic settings. The model supervision and alignment is done in English and the same model is used for reporting metrics across other languages like Basque, Italian, and Spanish."
}