Discovering and tracking organizational structures in event logs

The goal of process mining is to extract process-related information by observing events recorded in event logs. An event is an activity initiated or completed by a resource at a certain time point. Organizational mining is a subfield of process mining that focuses on the organizational perspective of a business process. It considers the resource attribute and derives a profile that characterizes the behavior of a resource in a specific business process. By relating resources associated with correlated profiles, it is possible to define a social network. This paper focuses on the idea of performing organizational mining of event logs via social network mining. It presents a framework that resorts to a stream representation of an event log. It adapts the time-based window model to process this stream, so that window-based social resource networks can be constructed, in order to represent interactions between resources operating at the data window level. Finally, it integrates specific algorithms, in order to discover (overlapping) communities of resources and track the evolution of these communities over consecutive windows. This paper applies the defined framework to two real event logs.