Mining Usage Scenarios in Business Processes: Outlier-Aware Discovery and Run-Time Prediction

A prominent goal of process mining is to build automatically a model explaining all theepisodes recorded in the log of some transactional system. Whenever the process to be minedis complex and highly-flexible, however, equipping all the traces with just one model mightlead to mixing different usage scenarios, thereby resulting in a spaghetti-like processdescription. This is, in fact, often circumvented by preliminarily applying clustering methodson the process log in order to identify all its hidden variants. In this paper, two relevantproblems that arise in the context of applying such methods are addressed, which havereceived little attention so far: (i) making the clustering aware of outlier traces, and (ii) findingpredictive models for clustering results.The first issue impacts on the effectiveness of clustering algorithms, which can indeed be led toconfuse real process variants with exceptional behavior or malfunctions. The second issueinstead concerns the opportunity of predicting the behavioral class of future process instances,by taking advantage of context-dependent "non-structural" data (e.g., activity executors,parameter values). The paper formalizes and analyzes these two issues and illustrates variousmining algorithms to face them. All the algorithms have been implemented and integrated intoa system prototype, which has been thoroughly validated over two real-life applicationscenarios.