A hybrid evolutionary-based process mining technology to discover parallelism structures

Information systems have been widely used to support workflow processes to record the execution of tasks in the process and are stored in so-called “event logs”. Techniques that relate to events extraction have gotten increasing attention such as process mining techniques. Developed process mining methods such as alpha algorithm, alpha⁺⁺ algorithm, and genetic process mining (GPM) are capable of tackling several structures well, but they are still difficult to discover parallelism structures efficiently since the parallelism structures are too complex. This work presents an evolutionary-based process mining approach based on a hybrid of GPM and particle swarm optimization algorithm (PSO) in order to handle parallelism structures. The medical records of acute stroke patients of Taiwanese medical institution are used as a practical case to test the proposed approach. Experimental results on the case show the effectiveness of the proposed approach for tackling parallelism structures.