Find in Library

Process mining aims to build the models of business processes and get valuable information according to event logs generated from enterprise information systems. There exist some indirect dependences, which refer to the relationship between discontinuous activities in business processes. However, the existing approaches cannot accurately identify such dependences from the event logs. Thus, this paper extends the <inline-formula> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula> algorithm and proposes a new one named the <inline-formula> <tex-math notation="LaTeX">$\alpha ^{\mathrm {TR}}$ </tex-math></inline-formula> algorithm, which uses the association rules to describe the indirect dependences. First, an algorithm is proposed to identify the choice and loop structures in the business process. Then, the association rules are mined to describe the indirect dependences. Finally, we design an extended Petri net to formalize the process model, which can accurately describe the indirect dependences. The effectiveness of the proposed approach is illustrated by the experiments on ProM.