Root-cause analysis of business processes: How humans utilize multiple sources of information to explain observations

Root-cause analysis of business processes seeks explanations and solutions to observed behaviors and problems in organizational business processes. Such analysis is usually based on event logs, utilizing process mining techniques. However, event logs hold a limited set of data attributes, and the analysis depends on data availability. To overcome this dependency, event log data can be complemented from additional sources that are commonly available in organizations. The aim of this research is to investigate how humans utilize potential combinations of event logs, databases, and transaction logs to explain observations. In particular, we conducted an empirical study, involving 73 participants, in order to: (1) find how these information sources and their combinations are used for answering questions related to violation of business rules; (2) identify composite operations that are performed when combining the information sources; and (3) gain insights into the perceived usefulness and usability of these combinations. Our findings provide evidence of the dominance of databases and event logs as the main sources of information. We further succeeded to classify typical composite operations into organizational information extension, behavioral information extension/refinement, single-source manipulation, and multi-source manipulation. Finally, these findings call for further support in process analysis and mining environments to improve usefulness and usability of multi-source root-cause analysis.