Efficient Discrete Particle Swarm Optimization Algorithm for Process Mining from Event Logs

Abstract Process mining, which aims to mine a high-quality process model from event log, provides powerful tool support the design, enactment, management, and analysis of operational business processes. However, task is not easy because algorithm needs discover various complex structures, handle noisy incomplete logs balance multiple performance indicators. In this paper, novel (called PSOMiner) for mining proposed, consists discrete particle swarm optimization guided local mutation. The former in charge searching solution space causal matrix latter used help skip out optimum when it suffers premature. A fine-grained scoring strategy assign score each position (i.e. matrix) presented direct experiments were performed on 28 synthetic with/without noise 4 real-life logs, three classical algorithms (ETM, Hybrid ILP Miner, HM) chosen comparison. results show that (1) PSOMiner achieved best f-score 25 logs; (2) average 0.825 superior ETM whose 0.703.