Complexity reduction of business process models with a cluttered "spaghetti" structure. (R-3250)

Understanding its business processes can be crucial for a company to attain a sustainable competitive advantage. When companies do not know the structure of their business processes, business process mining allows them to reconstruct the processes from event logs recorded by business information systems. The last decade, various process mining algorithms have been developed, mainly within a theoretical and technical context. The research literature also shows that these techniques face new challenges in a real business context. Firstly, the existing process mining techniques often generate extremely complex "spaghetti" models when applied to business data. Several approaches have recently been suggested, but further research on complexity reduction is needed. Secondly, existing process mining algorithms typically use a limited set of the available process data, which makes them less interesting for business experts. Overcoming these challenges is necessary for business process mining to reach its full potential in a business context. This project focuses on complexity reduction of process models learned from real business data using different perspectives, by (1) developing a sound complexity measure, (2) developing a trace clustering algorithm directly guided by model complexity and (3) developing a real-time process mining algorithm which detects changes in business processes over time.

Keywords:processmodeling

Disciplines:Mathematical sciences and statistics, Economics and business