A Multi-Aspect Semi-Automated Service Identification Method

Several service identification methods have been proposed to identify services using a business process-based strategy. However, these methods are still not accurate enough and adequately automated and thus need improvements. The present study addresses this gap by proposing a new semi-automated combinational method that applies process mining techniques and simultaneously considers different aspects of the business domain (e.g., goal and data). We argue that this method facilitates service identification more comprehensively and accurately and helps enhance organizational performance and lower cost structure. Our method includes three Phases. In the first phase, the system log is inspected, and the running business process is extracted using process mining techniques. After validating this model, we create a goal and data model in the next phase. In the third phase, we establish connections between the introduced models by defining some matrices. These connections are of two types: structural and conceptual. Finally, we propose a couple of algorithms that lead to the identification of services. We evaluate the utility of our proposed method by conducting a case study and using the experts’ opinions from different perspectives as follows: (1) assessing the accuracy and reusability of the identified services, (2) appraising the efficiency of employing this method in more complex processes, (3) calculating the cohesion to coupling ratio, and (4) assessing the performance of the method and other service quality measures. The results indicate that the average accuracy of this method is about 12 % higher than the previously identified methods for both simple and complex processes. Additionally, it empirically proves that using the process mining techniques improves the service identification considerably (8%). Moreover, according to the experts’ opinions, the combination of goal and data model and process mining has increased the performance by 8%. In comparison, the cohesion to coupling ratio demonstrated a 7% increase compared to other methods. In sum, we conclude that this method is an advanced and reliable way of service identification regardless of the process size and the complexity. The findings reveal that considering different aspects of business processes together and using process mining techniques improves the ratio of cohesion to coupling and accuracy of the identified services. Adherence to this approach enables companies to mine their business processes, modify them, and quickly identify services with higher performance. Besides, using this method provides a semi-automated and more effective way of service identification