Process Discovery Framework for Big Event Logs using Apache Spark

Nowadays significant volumes of data including event data are being collected and stored by various organizations and corporations. Discovering and extracting a process model from event data is the main objective of process discovery algorithms. Until now several different process discovery techniques have been proposed which are neither scalable nor based on big data architecture, and hence are unable to process hundreds of millions of events or activities. The present paper presents a process discovery framework based on map-reduce architecture using Apache Spark. Results of testing the proposed algorithm on two event logs with 30 million and 100 million traces respectively showed that the proposed framework outperforms existing tools such as Disco and MINIT