Event log filtering using graph clustering
Project Leader: Raffaele Conforti
Primary Contact: Raffaele Conforti (firstname.lastname@example.org)
Keywords: business process management; data mining; process mining; process modelling
Disciplines: Computing and Information Systems
In recent years, the number of companies opening up to the use of process mining is increasing. Using event logs as input, process mining allows companies to acquire insights on how to improve internal business processes. When the event log used as input is affected by noise the insights derived from a process mining exercise can be highly unreliable. Recent work has shown that chaotic activities, i.e. activities that can occur at any point in time during the execution of a process, have a negative impact on the quality of a discovered process model and that their removal positively impacts the precision of the model. This project aims at exploring the use of graph clustering techniques to identify and remove chaotic and/or infrequent activities from an event log to achieve better process discovery.