Persuasion Dialogues via Restricted Interfaces Using Probabilistic Argumentation

For persuasion dialogues between a software system and user, a user should be able to present arguments. Unfortunately, this would involve natural language processing which is not viable for this task in the short-term. A compromise is to allow the system to present potential counterarguments to the user, and the user expresses his/her degree of belief in each of them. In this paper, we present a protocol for persuasion that supports this type of move, and show how the system can use the epistemic approach to probabilistic argumentation to model the user, and thereby optimize the choice of moves.