EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

4326. Exploring Explainable Predictive Process Monitoring

Invited abstract in session MC-27: XAI in Business Processes, stream Mathematical Optimization for XAI.

Monday, 12:30-14:00
Room: 047 (building: 208)

Authors (first author is the speaker)

1. Andrei Buliga
Computer Science, Free University of Bozen-Bolzano

Abstract

This project focuses on advancing the field of Explainable Predictive Process Monitoring (XPPM) by developing novel techniques for generating explanations in PPM, and developing novel evaluation metrics. Furthermore, we aim to understand the interplay between different types of explanations and how they can be leveraged to improve the explainability process.

The first aspect of this research concerns the generation of counterfactual explanations in PPM, which are explanations that suggest what should be different in the input instance to change the outcome of a PPM model. Firstly, an evaluation protocol measuring the quality of the generated counterfactual examples and the evaluated method's performance is proposed.
In the second part, a methodology is proposed, that involves the use of genetic algorithms, incorporating temporal background knowledge at runtime. The methodology aims to generate counterfactual explanations that guarantee the satisfaction of the background knowledge, a critical aspect of counterfactual generation in XPPM.
A second aspect of this research is developing novel explainability method by using previously proposed Business Process Management (BPM) techniques to adapt explanation generation mechanisms.
A third aspect of this research project endeavours to combine various explanation techniques to streamline the search process and improve the overall quality of explanations, by leveraging multiple explanation types proposed in the literature.

Keywords

Status: accepted

Back to the list of papers