Mining Invisible Tasks in Non-free-choice Constructs

The discovery of process models from event logs (i.e. process mining) has emerged as one of the crucial challenges for enabling the continuous support in the life-cycle of a process-aware information system. However, in a decade of process discovery research, the relevant algorithms are known to have strong limitations in several dimensions. Invisible task and non-free-choice construct are two important special structures in a process model. Mining invisible tasks involved in non-free-choice constructs is still one significant challenge. In this paper, we propose an algorithm named alpha($). By introducing new ordering relations between tasks, alpha($) is able to solve this problem. alpha($) has been implemented as a plug-in of ProM. The experimental results show that it indeed significantly improves existing process mining techniques.