Big Data Meets Process Science: Distributed Mining of MP-Declare Process Models

Abstract

Process mining techniques allow the user to build a process model representing the process behavior as recorded in the logs. Standard process discovery techniques produce as output a procedural process model. Recently, several approaches have been developed to extract declarative process models from logs and have been proven to be more suitable to analyze flexible processes, which frequently depend on human decisions and are less predictable. However, when analyzing declarative constraints from other perspective than the control flow, such as data and resources, existing process mining techniques turned out to be inefficient. Thus, computational performance remains a key challenge of declarative process discovery. In this paper, we present a high-performance approach for the discovery of multi-perspective declarative process models that is built upon the distributed big data processing method MapReduce. Compared to recent work we provide an in-depth analysis of an implementation approach based on Hadoop, a powerful BigData-Framework, and describe detailed information on the implemented prototype. We evaluated effectiveness and efficiency of the approach on real-life event logs.