Process Mining for Unstructured Data: Challenges and Research Directions

The volume of data is continuously increasing and the ability and demand to efficiently analyze the data has become even more crucial. Machine learning and data mining are suitable techniques and tools to efficiently process and analyze the data. Complementary to both techniques is process mining [Aa16]. Process mining is a promising approach to find additional patterns (e.g., in terms of causal effects or bottlenecks) in data and in that way to give new insights into the data that could not be directly found with techniques like machine learning or data mining. The insights from processes are given by means of events that have been tracked by information systems. Then, this event data that is structured within a log (i.e., an event log), is used as input to any process mining algorithm. Process mining allows both an analysis based solely on event logs as well as a comparison between (manually generated or as-is) process models and an event log reflecting the to-be processes.