Petri Net of Thoughts: A Structure-Enhanced Prompting Approach for Process-Aware Artificial Intelligence

Prompt engineering, with techniques such as Chain/Tree/Graph-of-Thoughts, has emerged as a key method to enhance the reasoning capabilities of Large Language Models (LLMs) by guiding them through sequential and multi-path problem-solving approaches. In this paper, we introduce Petri Net of Thoughts (PNoT), a novel structure-enhanced prompting paradigm that leverages process discovery techniques to guide the reasoning process of an LLM during the generation of a Petri net model. PNoT represents the sequence, concurrency, and decision-making inherent in complex tasks. PNoT aims to improve both efficiency and interpretability in structured LLM-based reasoning tasks within the business process management domain.