Team Bruce Banner at Touché 2022: Argument Retrieval for Controversial Questions

• Tipo: Artigo em Livro de Atas de Conferência Internacional
• Data: 2022

Autores

Bernardo C. Moreira (Autor) FEUP; Henrique Lopes Cardoso (Autor) FEUP; Bruno Martins (Autor) Outra; Fábio Goularte (Autor) Outra

Ata de Conferência Internacional

Título: Proceedings of the Working Notes of CLEF 2022 - Conference and Labs of the Evaluation Forum, Bologna, Italy, September 5th - to - 8th, 2022.

Páginas: 3055-3063

2022 Conference and Labs of the Evaluation Forum, CLEF 2022

5 September 2022 through 8 September 2022

Indexação

Scopus - 1 Citação

Outras Informações

ID Authenticus: P-00X-5GB

Abstract (EN): Argument retrieval is a prominent topic in the context of current natural language processing applications. The task focuses on creating models that can retrieve coherent and strong arguments from textual sources. This technology can help individuals build an informed opinion about a controversial topic or support a particular stance on a debate. In this context, the Touché Task 1 was proposed within the scope of Conference and Labs of the Evaluation Forum 2022 (CLEF 2022), based on argument retrieval for controversial questions. We chose to compete with a sparse search. Despite ranking 3rd on relevance, 5th on quality and 4th on coherence, we concluded that our results are limited by our data arrangement process. The purpose of the task was to retrieve the most argumentative and relevant pairs of sentences, which could be formed with sentences from the same argument or not. Our approach focused on forming sentence pairs from the same argument, and achieved scores of 0.772 for quality, 0.651 for relevance, and 0.378 for coherence. © 2022 Copyright for this paper by its authors.

Idioma: Inglês

Tipo (Avaliação Docente): Científica