Argumentation provides a structured way to model and reason about conflicting opinions, allowing different viewpoints to be represented and evaluated in a principled manner. To solve computational problems in argumentation, existing solvers often rely on SAT (Boolean Satisfiability) or ASP (Answer Set Programming) techniques, with established reductions that provide effective ways of translating argumentation problems into SAT or ASP. Recent works have introduced structure-guided reductions that take the structure of the argumentation framework into account and can be particularly effective for counting problems.
In this thesis, you will implement the existing structure-guided reductions (for treewidth) for argumentation problems and systematically benchmark them against existing reductions that do not explicitly exploit the structure of the argumentation framework. The goal is to investigate whether incorporating structural information can improve solver performance, particularly for counting problems, and to identify the instances and settings where these reductions provide the greatest benefit. Possible tasks include implementing the reductions, integrating them with existing SAT-solvers, designing an experimental evaluation, and analyzing the results across different benchmark instances.