[This topic has already been assigned to a student.]

Description

Computer Game Playing has been a testbed since the early days of AI. During the last decades, much progress was achieved, where perhaps two of the most outstanding successes were IBM's Deep Blue beating Kasparov in 1997 and Checkers being solved by Jonathan Schaeffer's group in 2007. However, these game playing systems are of limited value for at least two reasons. For one, they are special-purpose programs that are only able to play the one particular game which they were created for. For another, such specialized systems usually contain a great deal of domain knowledge (e.g. in the form of heuristics), meaning that a large fraction of the game analysis is already done beforehand by human experts and designers.

[This topic has already been assigned to a student.]

Description

In a recent paper [1] we provided the theoretical foundation for the realization of a knowledge-based system based on the logic ES [2,3], a modal variant of the Situation Calculus [4,5]. The main advantages over existing approaches [6] is that it allows a more powerful query language, including nested belief operators and quantifying-in, and does not have to rely on meta-theoretic mechanisms to represent the agent's lack of knowledge.

The Knowledge-based Systems Group is looking for two student workers for software development and maintenance within the PLATAS project funded by the Deutsche Forschungsgemeinschaft (DFG).

Here's a list of theses supervised and finished at Knowledge-Based Systems Group.