An experimental evaluation of feedback loops in a business process mining genetic algorithm

This paper experiments with the use of feedback loops in a genetic business process mining algorithm. The use of genetic algorithms for process mining is explained along with a description of the research background to process mining. Of particular interest in this paper is the crossover operator. Experiments are described where problems encountered in mining processes are fed back into the crossover operator and used in the selection of crossover points. Both roulette wheel and tournament methods are used in the process of selecting crossover points. The paper concludes that the use of such problem feedback loops can be beneficial in the mining of simple business processes. However the paper makes clear that feedback loops are best employed as part of an ´intelligent´ mining technique.