Detecting Difference between Process Models using Edge Network

Process difference detection has played an important role in business process management for enterprise applications. However, the business processes are becoming more and more complex, involving a wide spectrum of tasks and different types of execution orders among these tasks, such as sequential, parallel, loop and conditional order. Thus, there is a need to detect difference between two process models efficiently. To meet this requirement, we use a difference detection framework for process models based on edge computing, where the edges can perform the task of difference detection between two process models, and the difference detection results can be aggregated to the cloud center. Most existing approaches detect process difference based on one feature of a process model, while a process model actually contains multiple features such as structure, behavior, and performance. In this paper, we propose an approach that can detect both structural and behavioral differences between two process models, which provides two aspects of difference information to process analysts and these kinds of insights are helpful to improve the original process model with low-cost and high-efficiency. First, we transform the process models into their corresponding task-based process structure trees (TPSTs) and assign each TPST node a feature vector based on the one-hot encoding. Then, the common key structure of two process models is extracted by comparing the feature vectors of nodes. Finally, the structural and behavioral differences are displayed in terms of this common key structure. Both the case study and efficiency study are provided to show the practicality of the proposed approach.