Argument Mining Pipeline
Argument mining requires several subtasks and often papers that are about argument mining focus on one of the tasks. However, if you were to implement a whole argument mining system, you would typically use the following pipeline with two main components, given raw text input:
- Argument component detection
- Argumentative sentence detection
- Argument component boundary detection
- Argument structure prediction
- Both intra- and inter-argument.
This structure was already used in the first (I think) argument mining paper, namely Mochales (2011) .
Lippie (2016) has the following graphical visualization: