Katalog Plus
Bibliothek der Frankfurt UAS
Bald neuer Katalog: sichern Sie sich schon vorab Ihre persönlichen Merklisten im Nutzerkonto: Anleitung.
Dieses Ergebnis aus BASE kann Gästen nicht angezeigt werden.  Login für vollen Zugriff.

NLP@IIMAS-CLTL at Multilingual Counterspeech Generation:Combating Hate Speech Using Contextualized Knowledge Graph Representations and LLMs

Title: NLP@IIMAS-CLTL at Multilingual Counterspeech Generation:Combating Hate Speech Using Contextualized Knowledge Graph Representations and LLMs
Authors: Márquez, David Salvador Preciado; Adorno, Helena Gómez; Markov, Ilia; Santamaria, Selene Baez
Contributors: Zubiaga, Irune; Montejo-Raez, Arturo; Soroa, Aitor; Martin-Valdivia, Maria Teresa; Guerini, Marco; Agerri, Rodrigo; Bonaldi, Helena; Vallecillo-Rodriguez, Maria Estrella
Source: Márquez, D S P, Adorno, H G, Markov, I & Santamaria, S B 2025, NLP@IIMAS-CLTL at Multilingual Counterspeech Generation : Combating Hate Speech Using Contextualized Knowledge Graph Representations and LLMs. in I Zubiaga, A Montejo-Raez, A Soroa, M T Martin-Valdivia, M Guerini, R Agerri, H Bonaldi, H Bonaldi & M E Vallecillo-Rodriguez (eds), Proceedings of the First Workshop on Multilingual Counterspeech Generation. Association for Computational Linguistics (ACL), pp. 29-36, 1st Workshop on Multilingual Counterspeech Generation, MCG 2025 with Shared Task on Multilingual Counterspeech Generation, Abu Dhabi, United Arab Emirates, 19/01/25. < https://aclanthology.org/2025.mcg-1.4/ >
Publisher Information: Association for Computational Linguistics (ACL)
Publication Year: 2025
Description: We present our approach for the shared task on Multilingual Counterspeech Generation (MCG) to counteract hate speech (HS) in Spanish, English, Basque, and Italian. To accomplish this, we followed two different strategies: 1) a graph-based generative model that encodes graph representations of knowledge related to hate speech, and 2) leveraging prompts for a large language model (LLM), specifically GPT-4o. We find that our graph-based approach tends to perform better in terms of traditional evaluation metrics (i.e., RougeL, BLEU, BERTScore), while the JudgeLM evaluation employed in the shared task favors the counter-narratives generated by the LLM-based approach, which was ranked second for English and third for Spanish on the leaderboard.
Document Type: conference object
Language: English
ISBN: 979-88-917-6207-7
Relation: info:eu-repo/semantics/altIdentifier/hdl/https://hdl.handle.net/1871.1/5ef8f0a4-6827-43d6-92d2-2eaaa1a3cdeb; info:eu-repo/semantics/altIdentifier/isbn/9798891762077; urn:ISBN:9798891762077
Availability: https://research.vu.nl/en/publications/5ef8f0a4-6827-43d6-92d2-2eaaa1a3cdeb; https://hdl.handle.net/1871.1/5ef8f0a4-6827-43d6-92d2-2eaaa1a3cdeb; https://www.scopus.com/pages/publications/105000137670; https://www.scopus.com/pages/publications/105000137670#tab=citedBy; https://aclanthology.org/2025.mcg-1.4/
Rights: info:eu-repo/semantics/openAccess ; http://creativecommons.org/licenses/by-nc-sa/4.0/
Accession Number: edsbas.8CB3D9DF
Database: BASE