A Time-Series Similarity Measure for Case-Based Deviation Management to Support Flexible Workflow Execution

Abstract

Our objective is to develop an approach based on case-based reasoning that detects and handles unforeseen deviations that occur in flexible workflow execution. With a case-based approach we aim at supporting the continuation of a deviant workflow execution by utilizing successfully completed processes, where similar deviations emerged. As a first step, this work introduces a novel similarity measure based on time sequence similarity that is able to compare running and completed workflow instances. We implemented and evaluated our approach in the ProCAKE framework. The proposed similarity measure achieves promising results considering runtime and similarity assessment.