Active Lable Based Technique for Business Process Model Abstraction

Abstract

Business process models are increasingly become an integral part for the organizations, to process their every operation in the form of models. Every organization’s stakeholder has a different perspective on a process and demands the appropriate model. As a matter of fact, there are multiple process models that capture the same business process. It is mainly observed that process repositories of organizations become very large because a process model is generated against every operation. Keeping such models in synchronize form is difficult in any changing business environment. To avoid this problem aggregation concept was introduced, as the Process model aggregation has the readable and high-level view of process model by aggregating and eliminating activities. Many researchers have been working on aggregation of process models and introducing different frameworks or approaches. Their research has concentrated on specific use cases and model transformation using these cases. But the problem we are facing today is that there is no such approach which is working on proper labeling style of activities for aggregation. As the result, poor quality of abstracted labels is generated. This thesis systematically approaches the problem of business process model aggregation in order to analyze the emerging need for aggregation in the industry. We develop model transformations that focuses on the semantics of process model elements are investigated. We propose a method for discovering semantically relevant events in process models. This research validates the function based aggregation method against sets of university process models and discuss the method implementation aspects. For this purpose, we will generate an algorithm to automate this aggregation process, apply precision, and recall as an evaluation measure.