An uncertainty-aware event log of network traffic

Business Process Management (BPM) heavily relies on event logs for process mining. However, traditional event logs may not always be available or may be harder to obtain for unlogged or unconventionally logged activities. To overcome these limitations, network traffic data can be used as an alternative source for constructing event logs. However, incorporating network traffic data poses its own set of challenges. These challenges include dealing with the large volume and diverse nature of network packets, as well as the uncertainty in mapping low-level events in a stream to specific activity types and border points, namely, the start and the end of an activity. In this paper, we introduce novel datasets that have been constructed from an enterprise network simulation environment. These datasets consist of two types of event logs: network traffic-level event logs and abstracted business-level event logs. Both types of logs exhibit various forms of uncertainty. These labeled datasets can serve as valuable benchmarks for a range of process mining tasks, such as event abstraction, process discovery, and conformance checking from uncertain event data.