Anomaly Detection based on Control-flow Pattern of Parallel Business Processes

Abstract

The purpose of this paper was to discover an anomalous-free business process model from event logs. The process discovery was conducted using a graph database, specifically using Neo4J tool involving trace clustering and data filtering processes. We also developed a control-flow pattern to address, AND relation between activities named parallel business process. The result showed that the proposed method improved the precision value of the generated business process model from 0.64 to 0.81 compared to the existing algorithm. The better outcome is constructed by applying trace clustering and data filtering to remove the anomaly on the event log as well as discovering parallel (AND) relation between activities.