Complexity-sensitive decision procedures for abstract argumentation

摘要: Abstract argumentation frameworks (AFs) provide the basis for various reason- ing problems in area of Artificial Intelligence. Efficient evaluation AFs has thus been identified as an important research challenge. So far, implemented sys- tems evaluating have either followed a straight-forward reduction-based approach or limited to certain tractable classes AFs. In this work, we present generic reasoning over AFs, based on novel concept complexity-sensitivity. Establishing theoretical foundations approach, derive several new complexity results preferred, semi-stable and stage se- mantics which complement current landscape abstract argu- mentation, providing further understanding sources intractability AF problems. The introduced framework exploits decision proce- dures lower whenever possible. This allows, par- ticular, instantiations via harnessing iterative way sophisticated Boolean satisfiability (SAT) solver technology solving considered First experimental show that SAT-based instantiation our outperforms existing systems.