Filtering Spurious Events from Event Streams of Business Processes Sebastiaan J. van Zelst 1 , Mohammadreza Fani Sani 2 , Alireza Ostovar 3 , Raffaele Conforti 4,⋆ , and Marcello La Rosa 4,⋆ 1 Eindhoven University of Technology, Eindhoven, The Netherlands s.j.v.zelst@tue.nl 2 RWTH Aachen University, Aachen, Germany fanisanit@pads.rwth-aachen.de 3 Queensland University of Technology, Brisbane, Australia alireza.ostovar@qut.edu.au 4 University of Melbourne, Melbourne, Australia {raffaele.conforti,marcello.larosa}@unimelb.edu.au Abstract. Process mining aims at gaining insights into business processes by analysing event data recorded during process execution. The majority of ex- isting process mining techniques works offline, i.e. using static, historical data stored in event logs. Recently, the notion of online process mining has emerged, whereby techniques are applied on live event streams, as process executions un- fold. Analysing event streams allows us to gain instant insights into business pro- cesses. However, current techniques assume the input stream to be completely free of noise and other anomalous behaviour. Hence, applying these techniques on real data leads to results of inferior quality. In this paper, we propose an event processor that enables us to effectively filter out spurious events from a live event stream. Our experiments show that we are able to effectively filter out spurious events from the input stream and, as such, enhance online process mining results. Keywords: Process mining, event stream, filtering, anomaly detection. 1 Introduction Nowadays, information systems can accurately record the execution of the business processes they support. Common examples include order-to-cash and procure-to-pay processes, which are tracked by ERP systems. Process mining [1] aims at turning such event data into valuable, actionable knowledge, so that process performance or compliance issues can be identified and rectified. Different process mining techniques are available. These include techniques for automated process discovery, conformance checking, performance mining and process variant analysis. For example, in process discovery we aim at reconstructing the underlying structure of the business process in the form of a process model, while in conformance checking we assess to what degree the recorded data aligns with a normative process model available in the organisation. The vast majority of process mining techniques are defined in an offline setting, i.e. they work over historical data of completed process executions (e.g. over all orders fulfilled in the past six months). They are typically not adequate to directly work in online settings, i.e. from live streams of events rather than historical data. Hence, they ⋆ Part of the work was done while the author was at the Queensland University of Technology.