Extending LIME for Business Process Automation

Abstract

AI business process applications automate high-stakes business decisions where there is an increasing demand to justify or explain the rationale behind algorithmic decisions. Business process applications have ordering or constraints on tasks and feature values that cause lightweight, model-agnostic, existing explanation methods like LIME to fail. In response, we propose a local explanation framework extending LIME for explaining AI business process applications. Empirical evaluation of our extension underscores the advantage of our approach in the business process setting.

References in corpus (6)

• European Union regulations on algorithmic decision-making and a "right to explanation"
• Predictive Business Process Monitoring with LSTM Neural Networks
• Predicting Process Behaviour using Deep Learning
• Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods
• Problems with Shapley-value-based explanations as feature importance measures
• Model Agnostic Multilevel Explanations

Cited by in corpus (2)

• How well can a large language model explain business processes as perceived by users?
• SMACE: A New Method for the Interpretability of Composite Decision Systems