Mining Collaboration Business Process Containing Invisible Task by Using Modified Alpha

Abstract

Business processes are experiencing increasingly complex developments; therefore, an extensive business process must cover all existing process flows. Applied business process collaboration between organizations can complete a complex business process. The additional information which shows the collaboration of activities is called messages. Process discovery is currently focused on a series of activities in a single process model, so the process discovery cannot depict the messages in the business process collaboration. In addition, there are several problems in describing the condition of activities, e.g., an Invisible Task. The Invisible Task is a condition of additional tasks that appear not in the event logs but in the process models. The Invisible Task must be described in the process model; therefore, it can be analyzed further. Several conditions which need the Invisible Task are redo, switch, and skip conditions. In this research, the proposed method is to obtain information about the event log of all activities of the business process collaboration and discover any Invisible Task to describe in the process model. The proposed method, named the Modified Alpha algorithm, builds several rules for adding messages and the Invisible Task in the event log before executing the Alpha algorithm. The results of this study indicate that the Modified Alpha algorithm can describe the collaboration process model. Based on the comparison results, the Modified Alpha algorithm gets the best results than other algorithms, namely Alpha# and Inductive Miner. Modified Alpha received 1.00, 1.00, 1.00, 0.82 for the fitness, precision, simplicity, and generalization. Alpha# Miner earned 0.74 and 0.70 for the simplicity and generalization, and Inductive Miner gained 0.55 simplicity value and 0.72 generalization value. Alpha# Miner and Inductive Miner got 0.00 for the fitness and the precision.