Hatschka, C., Ciabattoni, A., & Eiter, T. (2022, November 28). Representing Normative Reasoning in Answer Set Programming Using Weak Constraints [Conference Presentation]. TAASP - Workshop on Trends and Applications of Answer Set Programming, Wien, Austria. http://hdl.handle.net/20.500.12708/154091
E192-01 - Forschungsbereich Algorithms and Complexity E192-03 - Forschungsbereich Knowledge Based Systems E192-05 - Forschungsbereich Theory and Logic
-
Date (published):
28-Nov-2022
-
Event name:
TAASP - Workshop on Trends and Applications of Answer Set Programming
-
Event date:
28-Nov-2022 - 29-Nov-2022
-
Event place:
Wien, Austria
-
Keywords:
answer set programming; normative systems
en
Additional information:
Deontic logics and normative reasoning keep gaining importance, as
ethical AI and multi-agent systems are becoming increasingly more relevant these
days, and tools that support decision making based on norms and regulations are
needed. Challenges to this are illustrated by well known benchmark examples
(referred to as deontic paradoxes) on which common deontic logics, e.g., Standard
Deontic Logic, fail. In this work we encode a range of famous paradoxes in
Answer Set Programming (ASP) using weak constraints. Their abstraction and
generalization provides a plain methodology for encoding normative systems in
this language. An experimental comparison to the normative supervisor by Neufeld
et al. (CADE 2021) on "ethical" Pacman shows that the ASP encoding leads to
comparable performance but ethically preferable results.
-
Research Areas:
Logic and Computation: 80% Mathematical and Algorithmic Foundations: 20%