In recent years, several techniques have been made available to automatically discover declarative process models from event logs. These techniques are useful to provide a comprehensible picture of the process as opposed to full specifications of process behavior provided by procedural modeling languages. Since many modern systems produce ``big data'' from business process executions, in previous work, a framework for the discovery of LTL-based declarative process models from streaming event data has been proposed. This framework can be used to process events online, as they occur, as a way to deal with large and complex collections of datasets that are impossible to store and process altogether. However, the proposed framework does not take into account data attributes associated with events in the log, which can otherwise provide valuable insights into the rules that govern the process. This paper makes the first proposal to close this gap by presenting a technique for discovering declarative process models from event streams that incorporates both control-flow dependencies and data conditions. Specifically, we use Hoeffding trees to incrementally discover data-aware declarative process models, which are represented as conjunctions of first-order temporal logic expressions. The proposed technique has been validated on a synthetic event log, and on a real-life log of a cancer treatment process.
Published in Proceedings of the International Joint Conference on Neural Networks (IEEE WCCI IJCNN 2020); Glasgow, Scotland, UK; July 19-24, 2020.