PhD Thesis

The project is funding two PhD thesis.

• Juliete Rossie (2023-2026).
• Jordan Thieyre (2023-2026).
  Title: Computational models of arguments for online debates
  Supervisors: Aurélie Beynier, Nicolas Maudet, Srdjan Vesic
  Institution: LIP6, Sorbonne Université, Paris
  
  Description: Online debate platforms are based on the idea that better collective decision-making may emerge thanks to articulated debate. However, these platforms remain typically only conceived as visualisation and argument mapping tools. There exists opportunities though to exploit formal argumentation theory to improve these tools and better capture the dynamics of these debates, for instance to assess the relevance of exchanged arguments, or avoid malicious behaviours. Our work is to investigate how formal argumentation theory and computational social choice can be exploited to help users and platform designers to better understand the debates and the arguments presented therein.

Engineering Contribution

The project is funding an engineering position.

• Theo Alkibiades Collias.

Master Internship

• Clément Lens (2023).
  Title: Exploring the nuances of voting in online argumentation debate systems
  Supervisors: Jérôme Delobelle, Sébastien Konieczny, Srdjan Vesic
  Institution: CRIL, Université d'Artois, Lens
  
  Description: As online argumentation systems become increasingly prevalent, the volume of arguments and options within these systems continues to grow. This presents a significant challenge for human analysts, who struggle to manually track changes, understand the various arguments and their relationships, and make sense of the vast amount of information. To address this challenge, our project proposes to use artificial intelligence and argumentation theory to study these systems. Specifically, we will focus on the problem of votes on arguments, which can be either positive or negative. Currently, there is no approach in the literature that can handle votes in an elaborated way. The existing proposal is to calculate the ratio of positive and negative votes and use this as a single number. Our goal is to explore more ways to take into account the votes and understand the nuances of the voting process. One important question that must be addressed is the interpretation of votes. Does voting for an argument mean that the person agrees with the argument's existence, its formal correctness, its hypothesis or conclusion, wants to believe that the argument is true, or simply wants others to believe it is true? The goal of this master thesis will be to determine if these different possible interpretations have to change the way the votes are taken into account in the analysis.
  
  
• Théophile Romieu (2025).
  Title: Study of Contribution Functions with respect to Gradual Semantics
  Supervisors: Elise Bonzon, Jérôme Delobelle, Julien Rossit
  Institution: LIPADE, Université Paris Cité
  
  Description: For gradual semantics to be used in practice, it is essential to provide explanations for the returned results. Among the available tools for generating explanations are contribution measures, whose purpose is to quantify the impact of an argument or a set of arguments on a given target argument. A set of principles has been proposed to axiomatically evaluate a contribution measure associated with a given gradual semantics. However, whenever a new semantics is introduced, a complete axiomatic study must be carried out again. In this work, the goal is not to focus on a specific semantics, but rather to rely on the properties satisfied (or not) by a gradual semantics in order to determine whether the principles related to contribution measures are also satisfied.