The paper presents an analysis of contribution functions for quantitative bipolar argumentation graphs. These functions quantify how one argument contributes to another. The analysis introduces principles that formalize the intuitions underlying different contribution functions, and expectations regarding their behavior. While no function satisfies all principles, the analysis serves as a tool for selecting the most suitable function based on the requirements of a given use case. The paper aims to enhance understanding of argument influence, reflecting ideas of feature attributions in machine learning and contributions of agents in cooperative game theory.
Publication date: 18 Jan 2024
Project Page: arXiv:2401.08879v1
Paper: https://arxiv.org/pdf/2401.08879
