Concepts of Conditional Independence in Computational Argumentation
Aus International Center for Computational Logic
Concepts of Conditional Independence in Computational Argumentation
Vortrag von Lydia Blümel
- Veranstaltungsort: APB-2026
- Beginn: 19. Februar 2026 um 11:00
- Ende: 19. Februar 2026 um 12:00
- Forschungsgruppe: Logische Programmierung und Argumentation
- Event series: Research Seminar Logic and AI
- iCal
The ability to recognize and manage (in)dependence is crucial in symbolic reasoning , it plays a major part in causal reasoning [Pearl, 2009], e.g., in probabilistic models, where (in)dependence between variables of interest is built into graphical representations such as Bayesian networks. The ability to determine independence contributes to the explainability and trustworthiness of reasoning systems, while also improving efficiency by empowering breaking problems down so that they can be managed more effectively. The speaker will present some recent work on notions of conditional independence in computational argumentation. Two prominent areas of computational models of argument are abstract argumentation [Dung 1995], and structured argumentation. Unlike abstract argumentation, where arguments are considered abstract, structured argumentation frameworks focus on the inner structure of the arguments, enabling a finer-grained analysis of the arguments and relations. Assumption-based argumentation (ABA) [Bondarenko et al., 1997] is a well-known form of structured argumentation, whose building blocks are assumptions (defeasible elements) and inference rules. In both frameworks argumentation semantics provide methods for evaluating acceptance, and it is wrt. argumentation semantics that [Rienstra et al., 2020] propose a notion of conditional independence for Abstract Argumentation, which recently was extended to independence between assumptions in Flat ABA-Frameworks [Blümel et al., 2025]. However, in both frameworks other concepts of conditional independence are conceivable, especially in ABA, where assumptions are not the only structural objects under investigation. Furthermore, the current notion has proven computationally challenging. In the second part of the talk the speaker will discuss some ideas regarding further investigations along these lines.
