Kontra-Narratiben Sorkuntza Automatikoa: Gorrotoa Deseraikiz

Abstract

This work introduces ML-MTCONAN-KN, the first knowledge-based multilingual corpus for counterargumentation. The corpus is structured in a parallel manner across four languages: English, Basque, Spanish, and Italian. This dataset represents a significant advancement in the field, addressing the scarcity of resources for counter-argumentation and marking the first multilingual collection of knowledge-based counternarratives. Additionally, we provide an in-depth analysis of automatic counter-narrative generation in Basque, a study that has not been conducted before. Our contribution aims to enhance the development of systems capable of generating constructive responses to hate speech. Content Warning: This paper contains hate speech that may be offensive to some readers. These views do not reflect those of the authors.