University of Twente Student Theses
Input Data Reduction on Natural Language Explanations of Business Processes using Large Language Models
Oerle, Patrick van (2025) Input Data Reduction on Natural Language Explanations of Business Processes using Large Language Models.
|
PDF
427kB |
| Abstract: | Process Mining can give valuable insights to businesses based on event log data to analyze and optimize business processes. With recent advancements such as OpenAI's GPT-4 and Meta's LLaMA, Large Language Models (LLMs) are now able to quickly and accurately analyze a provided business model and give suggestions to tackle inefficiencies in the model. However, such process models are constructed based on large amounts of data, making the use of an LLM time-consuming and resource-consuming. When some way can be found where a significant reduction of input data while an LLM is still able to give an accurate explanation of a process model, a lot of time and computational resources can be saved. During this research, Meta's LLaMA 3.3 will provide an explanation of a business model based on different amounts of input data, which is then given a score. The results show that after some amount of input data, using significantly more event logs only slightly increases the score given. This implies that significant data reduction for LLMs in process mining should be possible. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/105103 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page