Merging Event Logs for Inter-organizational Process Mining

Abstract

In an inter-organizational environment, the discovery of process choreography is challenging because the different organizations involved have to put together their partial knowledge about the overall collaborative business process. This chapter presents a methodology to merge the event logs of the different partners of a collaborative business process, in order to serve as input for the process mining algorithm. On the one hand, the methodology consists of a method set for searching the correlation between events of the log of different partners involved in the collaboration. These methods are implemented at the trace level and the activity level. On the other hand, the methodology consists of a set of methods and rules for discovering process choreography. From the knowledge gained by the above methods, message-type tasks are identified and marked in each event log, then using a formal set of rules, the message task sub-type (send or receive) is discovered. Finally, links using message sequence flow connectors between message tasks identified as pair activities in event logs are automatically defined. The proposed approach is tested using a real-life event log that confirms their effectiveness and efficiency in the automatic specification of message flows of the process choreography discovered, allowing to build a collaborative business process model.