Abschlussarbeit

The area of iterated belief change [1,2] investigates the evolution of belief change strategies of agents. Two main operations considered are revision and contraction. Revision is the kind of belief change where a new belief is incorporated into the initial beliefs such that the new information is prioritized and inconsistencies are solved whenever that is possible. Contraction is the process of withdrawing a belief from the agent's beliefs without adding new ones. In literature, there are several change strategies known for belief change. Prominent approaches [3] are natural change, which demands that the change strategy only evolve minimally, and lexicographic change (also called moderate), which requires that the change strategy is maximally adapted to new arriving information. For contraction, it was proposed [4] to "weaken" natural and lexicographic change to more gentle versions according to two principles:

• When no change is necessary on the beliefs, do not (drastically) change the change strategy.
• When a change is necessary for the beliefs, a change of the change strategy itself should also be possible.

The definition and characterization of gentle natural and lexicographic revision versions is an open research topic.

This open research topic allows different kinds of theses. Example topics are:

• Master's thesis: define and characterize gentle versions of natural and lexicographic revision in the spirit of gentle contraction strategies. This would involve intensive study of these kinds of operations by examples.
• Bachelor's or master's thesis: implementing the above-mentioned change strategies for contraction or revision.
• Bachelor's or master's thesis: the above-mentioned topics for gentle versions for other operations. Depending on personal interests and whether it shall be a master's or bachelor's thesis, a specific focus can be defined together with the student.

• [1] Eduardo L. Fermé, Sven Ove Hansson: Belief Change - Introduction and Overview. Springer Briefs in Intelligent Systems, Springer 2018
• [2] Adnan Darwiche, Judea Pearl: On the Logic of Iterated Belief Revision. Artif. Intell. 89(1-2), 1997, p. 1-29
• [3] Richard Booth, Thomas Andreas Meyer: Admissible and Restrained Revision. J. Artif. Intell. Res. 26, 2006, p. 127-151
• [4] Kai Sauerwald: Semantics of Belief Change Operators for Intelligent Agents: Iteration, Postulates, and Realizability. Dissertations in Artificial Intelligence 352, IOS Press, 2022