Big data quality: Early Detection of Errors in Process flow using Petri Nets with Data

Abstract

In our Big Data era, data is being produced at scale, in motion, and in heterogeneous forms. Uncertainty is another significant attribute exhibited by this data and hence there is need to comprehend and (perhaps) repair erroneous data timely. Due to heterogeneity of data source and usage, data quality rules are contextual; hence we require data management solutions that acknowledge these varied uses and incorporate them to determine the level of quality and standardization required. Today, there is a wide range of process mining techniques that are able to uncover the reality of processes through a systemic analysis of event data. These techniques are being applied in this work with the aim to isolate the source of the introduction of data flaws to fix the process instead of correcting the data. This paper employs the Heuristic Miner algorithm for process discovery, Petri nets with data (DPN nets) and conformance checking using alignments and compliance rules. We showed that alignments between event logs and the discovered Petri Net from process discovery algorithms reveal frequent occurring deviations and compliance rules are an effective data management solution. Insights into these deviations are then exploited to repair and enhance the original process models. Our novel diagnostic data-aware process discovery technique is applied on a real-life event log and evaluated for its success in providing new and valuable insights and failure in other areas of performance.