Can We Find Better Process Models? Process Model Improvement Using Motif-Based Graph Adaptation

摘要

In today's organizations efficient and reliable business processes have a high influence on success. Organizations spend high effort in analyzing processes to stay in front of the competition. However, in practice it is a huge challenge to find better processes based on process mining results due to the high complexity of the underlying model. This paper presents a novel approach which provides suggestions for redesigning business processes by using discovered as-is process models from event logs and apply motif-based graph adaptation. Motifs are graph patterns of small size, building the core blocks of graphs. Our approach uses the LoMbA algorithm, which takes a desired motif frequency distribution and adjusts the model to fit that distribution under the consideration of side constraints. The paper presents the underlying concepts, discusses how the motif distribution can be selected and shows the applicability using real-life event logs. Our results show that motif-based graph adaptation adjusts process graphs towards defined improvement goals.