Syamsiyah, Alifah, van Dongen, Boudewijn F. & Dijkman, Remco M. (2018). A native operator for process discovery. In Hui Ma, Günther Pernul, Abdelkader Hameurlain, Sven Hartmann & Roland R. Wagner (Eds.), Database and Expert Systems Applications – 29th International Conference, DEXA 2018, Proceedings (pp. 292-300). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), No. 11030 LNCS). Springer-Verlag Berlin Heidelberg.

Abstract

The goal of process mining is to gain insights into operational processes through the analysis of events recorded by information systems. Typically, this is done in three phases. Firstly, events are extracted from a data store into an event log. Secondly, an intermediate structure is built in memory and finally, this intermediate structure is converted into a process model or other analysis results. In this paper, we propose a native SQL operator for direct process discovery on relational databases. We merge steps 1 and 2 by defining a native operator for the simplest form of the intermediate structure, called the “directly follows relation”. We evaluate our work on big event data and the experimental results show that it performs faster than the state-of-the-art of database approaches.