Desktop Activity Mining - A new level of detail in mining business processes

Abstract

New analysis and automation technologies are significantly changing the way how business process management is performed. Especially Robotic ProcessAutomation (RPA) is rapidly gaining importance as a method to automate office processes. An efficient automation of office processes however requires detailed information about all user activities related to the process. While process mining techniques can in principle be used to discover processes in a data-driven way, the existing approaches are not able to gather information in a level of detail required for automation purposes. That is why in particular the configuration of RPA systems is a labor and knowledge-intensive task that is based on a human expert, modeling all process variations in detail. In this paper, we present Desktop Activity Mining as a new approach to mine detailed process activity data. The concept is to record the detailed desktop activities of all users performing an office process and consolidate the process variations with process mining techniques to discover an integrated process model. As a proof of concept, we realized a prototypical implementation. Our findings suggest that Desktop Activity Mining holds the potential to optimize not only process automation but also to derive a new level of detail in mining and analyzing business processes.