Log Shifting for Enterprise Collaboration Systems: A Supervised Event Abstraction Approach (Extended Abstract)

Event abstraction is a pre–processing method in Process Mining to create an interpretable log and, in turn, interpretable process models. Here, one or more events are aggregated to one event at a higher level of abstraction. The ProM plugin presented in this paper provides such an approach by training a log shifting model based on investigating a known high–level log and the related low–level log. This log shifting model can then be used to convert any low–level log of the same application. First evaluation results show that this approach is applicable. The log shifter presented in this paper is particularly tailored for Enterprise Collaboration Systems (ECS) that come with special low–level log characteristics.