Enhancing Process Mining through LLMs: A User-Centric Approach

The integration of large language models (LLMs) in process mining offers unprecedented opportunities for enhancing the efficiency and effectiveness of process mining utilization within organizations. Existing research on LLMs in process mining often overlooks the diverse needs of different user roles. This study addresses this gap by presenting a user-centric framework that demonstrates how LLMs can enhance process mining tasks for Operations Managers, Data Scientists, and Process Architects. The primary objective of this research is to develop a comprehensive framework that elucidates the capabilities of LLMs across these roles and demonstrates how to harness these capabilities optimally. To achieve this, we first analyze the specific contributions and responsibilities of each role in the context of process mining. We then identify how LLMs can assist in their respective tasks using the BPIC 2017 dataset as a case study. This approach highlights the unique ways LLMs can enhance the work of each user group, providing tailored insights and recommendations. Our methodology includes a thorough examination of role-specific tasks and the application of LLMs to improve task performance and outcomes. Our findings offer a novel perspective on the application of LLMs, presenting a user-centric framework that serves as a comprehensive guidebook for leveraging LLMs to their fullest potential.