Establishing Requirements for the Application of Process Discovery Algorithms in Audit Practice

Abstract

Process mining's flagship approach, process discovery, powers the reconstruction of a com-pany's processes as they transpired in practice, instead of reflecting anyone's expectations of their execution. This re-establishment is based on the company's full collection of transactions , as recorded (partly) independently of the auditee in his information systems. For audit practice to fully benefit from process discovery's potential, it should nevertheless be ascertained that the algorithms respond to audit concerns. This study explores the prerequisites that contemporary process discovery research imposes on successful discovery algorithms to ensure that they are both user-friendly and proficient in the expression of a process' behaviour. By drawing a parallel between these prerequisites and audit standards and expectations , a set of requirements is obtained that need to be satisfied by a discovery algorithm in order to be functional in an audit context. The practical use of the requirements for the evaluation of current and forthcoming process discovery algorithms on their value in audit practice is demonstrated for four state-of-the-art process discovery algorithms.