Discovering multi-perspective process models

Process Mining techniques exploit the information stored in the executions log of a process in order to extract some high-level process model, which can be used for both analysis and design tasks. Most of these techniques focus on "structural" (control-flow oriented) aspects of the process, in that they only consider what elementary activities were executed and in which ordering. In this way, any other "non-structural" information, usually kept in real log systems (e.g., activity executors, parameter values, and time-stamps), is completely disregarded, yet being a potential source of knowledge. In this paper, we overcome this limitation by proposing a novel approach for discovering process models, where the behavior of a process is characterized from both structural and non-structural viewpoints. In a nutshell, different variants of the process (classes) are recognized through a structural clustering approach, and represented with a collection of specific workflow models. Relevant correlations between these classes and non-structural properties are made explicit through a rule-based classification model, which can be exploited for both explanation and prediction purposes. Results on real-life application scenario evidence that the discovered models are often very accurate and capture important knowledge on the process behavior.