Exploiting Treewidth for Counting in Abstract Argumentation

Talk by Piotr Gorczyca

Location: Online
Start: 23. September 2021 at 1:00 pm
End: 23. September 2021 at 2:30 pm
Event series: Research Seminar Logic and AI
iCal

Abstract argumentation and Dung’s framework is popular for modeling and evaluating arguments in artificial intelligence. Recently, counting problems in abstract argumentation became of increasing interest visible by the 2021 ICCMA competition, which asked to solve counting problems despite the high computational complexity. In this paper, we consider various counting problems in abstract argumentation under practical aspects. We revisit previous theoretical algorithms, present formulations in relational algebra, and an implementation employing the DPDB framework for solving problems using dynamic programming on tree decompositions. Thereby, we establish an implementation for counting extensions of abstract argumentation frameworks under admissible, stable, and complete semantics. We provide a thorough empirical evaluation that also incorporates less known approaches to solve counting questions in argumentation. Finally, we show that our approach can be particularly well suited for instances of low treewidth and is quite useful in a portfolio-based solving approach.

This is a project presentation and with an approximate duration of 30 minutes and a 15 minutes Q&A session. It will be given online via BigBlueButton. To access the room, use one of the following links:

with ZIH-login:

https://selfservice.zih.tu-dresden.de/l/link.php?m=147897&p=76c02e6b

without ZIH-login:

https://selfservice.zih.tu-dresden.de/link.php?m=147897&p=05cc674a