Exploiting Large Language Models for the Automated Generation of Constraint Satisfaction Problems

Lothar Hotz

Abstract:

Constraint Satisfaction Problems (CSPs) are a core technology that solves many real-world problems, especially
for configuration tasks. A key success factor in this context is an efficient knowledge acquisition process where
domain experts and knowledge engineers (developers of CSPs) should develop an agreement on the correctness
of the expanding knowledge base as soon as possible. In this paper, we show how large language models
(LLMs) can be applied to the automated generation of solutions for constraint satisfaction problems thus
reducing overheads related to CSP development and maintenance in the future.