Mining workflow event log to facilitate parallel work item sharing among human resources

Abstract

In many workflow applications, actors are free to pick up work items in their work list. It is not unusual for an actor to start a work item before completing other previously accepted ones. Frequent occurrence of this behaviour implies potential patterns of work parallelism, which is useful for workflow scheduler to better dispatch ongoing work items. In this article, we apply association rule mining techniques to workflow event log to analyse various kinds of activity parallel execution patterns. When an actor accepts a new work item, the parallel execution rules mined from event log can help the workflow scheduler to find other work items that might be suitable to be undertaken by the same actor simultaneously. In the experiment on three vehicle manufacturing enterprises, we have found 32 strong rules of 40 different workflow activities. We describe our approach and report on the result of our experiment.