Abstract

Arguments are the fundamental building blocks of any argumentation system, serving as the primary means of supporting claims with evidence. However, in many real-world settings—such as dialogues, debates, and multi-agent interactions—arguments are rarely independent. They often rely on common evidence, or are put forward by different agents expressing closely related ideas in different ways. Such situations naturally give rise to various forms of similarity between arguments.

In this talk, I will address several fundamental questions concerning argument similarity. What does it mean for two arguments to be similar? How can similarity be formally defined and quantitatively measured? How should similarity influence the evaluation of an argument's strength or acceptability? Finally, how can argumentation semantics be extended to explicitly account for similarity relations between arguments?

To answer these questions, I will present a systematic overview of existing approaches to argument similarity. I will also highlight the key conceptual and technical differences between extension-based and gradual semantics, with particular emphasis on how these two paradigms incorporate—or fail to incorporate—similarity into the evaluation process.

Abstract
Preferential approaches to defeasible reasoning have turned out to be particularly promising for the propositional case, mainly because they are based on an elegant, comprehensive, and well-studied framework for non-monotonic reasoning proposed by Kraus, Lehmann, and Magidor—often referred to as the KLM approach. In the first part of this talk I will provide an introduction to propositional KLM-style defeasible reasoning. This will be followed by an overview of attempts to extend KLM-style defeasible reasoning beyond propositional logic. More specifically, I’ll look at the application of the KLM approach to description logics and some restricted first-order logics. In doing so, I’ll describe what has worked well, but also what are the (many) remaining challenges.

Abstract

Given a knowledge base, forgetting aims to reduce its signature while preserving the semantic relations over the remaining signature. In the context of Answer Set Programming (ASP), a wide variety of forgetting approaches has been proposed, often based on differing—or even conflicting—objectives and intuitions.

In this talk, we will discuss the main approaches to forgetting in ASP, focusing on the challenges and limitations of preserving all semantic relations over the remaining signature. We will also highlight connections to recent work on abstraction in ASP.