Active Anomaly Detection for Key Item Selection in Process Auditing

Abstract Process mining allows auditors to retrieve crucial information about transactions by analysing the process data of a client. We propose an approach that supports identification unusual or unexpected transactions, also referred as exceptions. These exceptions can be selected “key items”, meaning wants look further into underlying documentation transaction. The encodes traces, assigns anomaly score each trace, and uses domain knowledge update assigned scores through active detection. is evaluated with three groups over cycles. results evaluation indicate has potential support decision-making auditors. Although still need make manual selection key items, they are able better substantiate this selection. As such, our research seen step forward respect usage detection analysis in auditing.