| Title |
A Comparative Study of Explainable AI Techniques for Process Analysis |
| Authors |
정찬일(Chanyil Jung) ; 이후진(Hoojin Lee) |
| DOI |
https://doi.org/10.5573/ieie.2020.57.8.51 |
| Keywords |
Process mining; Process analysis automation; Machine learning; eXplainable AI |
| Abstract |
There have been several attempts to apply artificial intelligence to process mining, but the focus is on process prediction, not on the causal analysis, which is the core purpose of process mining. If we use Explainable AI (XAI) to find issues in business processes and their causes, we can analyze the causes of the issues regardless of the expert's personal capabilities. However, the XAI technology is still in development, and various methods are mixed, so that the methodology and algorithm most suitable for automating of the process causal analysis have not been organized. In this paper, we thus apply various machine learning models including deep learning and various XAI algorithms such as LIME, SHAP and LRP to process analysis automation, and compare the characteristics and advantages of each algorithm to find the optimal process analysis automation methodology. If the proposed machine learning technique and XAI algorithm are used, it will be possible to automatically and quantitatively and easily analyze the cause of business issues through process mining based on an algorithm. |