A Novel Process Mining Algorithm to Discover Non-free Choice Construct from Event Logs

Abstract

It is always a challenge in the field of process mining to mine the non-free choice construct combining choice and synchronization from the event log. To solve this problem, we propose an improved process mining algorithms based on the genetic process mining. In this paper, we present a new definition of the ordering relations that can determine long distance dependency, which can build the initial population more biasedly, and prepare sufficient high quality individuals for subsequent evolutionary calculations. Then we can reduce the search space and avoid the existence of inferior individual. Mining results are validated by fitness, precision and generalization. Experimental results show that the improved algorithm is better than the existing algorithm for mining non-free choice construct. These improvements make the mining of process model more accurately reflecting the business processes.