• In real-life, arguments do not come structured and their relations is not highlighted.
• They are usually hidden within dialogues or debates.
• While one could manually annotate them, it is a time-consuming task and researchers have tried to use automatic techniques to find them.
• In this lecture, we look at several ML techniques used to deal with several argumentation task.

6.1 Reminder on the BERT model

6.2 Fallacy detection in political debates (Goffredo et al. 2023)

6.3 Relation-based argument mining (RBAM)

6.4 BERT-based models for RBAM (Ruiz-Dolz et al. 2021)

6.5 RBAM on debates (Agarwal et al. 2022)

6.6 RBAM using LLMs and in-context learning (Gorur et al. 2024)

6.7 RBAM using LLMs and fine-tuning (Faugier et al. 2024)

6.8 Conclusion