A Node Type and Logs Combination-based Recommendations for Business Process Modeling

In today's ever-changing business environment, enterprises face numerous challenges, including rapid changes in customer needs, intense market competition, and swift technological advancements. Consequently, business process modeling has emerged as a pivotal strategy to tackle these challenges, optimize operations, and enhance efficiency. Traditionally, manual modeling required participants to possess domain knowledge and modeling skills, rendering the process highly complex and prone to errors. To address this, business process recommendation techniques have been introduced to effectively assist business process modeling by leveraging node information from business process repositories. However, existing research on process recommendation often overlooks crucial factors such as node-type information and fails to integrate actual execution log data, resulting in suboptimal recommendation results. In this study, we consider the node-type information contained in the process to design the edge expansion model and combine it with the log information generated during the actual execution of the process to calculate the confidence level comprehensively and as the basis for the recommendation to conduct offline mining, the results of mining to assist business process analysts in completing the process modeling. Furthermore, we perform a comparative analysis between our proposed algorithm and mainstream algorithms in process recommendation, utilizing a real dataset. The results demonstrate the superiority of our method in terms of precision and branch structure recommendation.