Since the environment for businesses is becoming more competitive
by the day, business organizations have to be more adaptive to
environmental changes and are constantly in a process of optimization.
Fundamental parts of these organizations are their business processes.
Discovering and understanding the actual execution flow of the processes
deployed in organizations is an important enabler for the management,
analysis, and optimization of both, the processes and the business. This
has become increasingly difficult since business processes are now often
dynamically changing and may produce hundreds of events per second.
The basis for this paper is the Constructs Competition Miner (CCM): A
divide-and-conquer algorithm which discovers block-structured processes
from event logs possibly consisting of exceptional behaviour. In this paper
we propose a set of modifications for the CCM to enable scalable
dynamic business process discovery of a run-time process model from
a stream of events. We describe the different modifications and carry
out an evaluation, investigating the behaviour of the algorithm on event
streams of dynamically changing processes.