Building process models based on interval logs

Abstract

Contemporary process management systems are driven by explicit process models, i.e., a workflow design is required in order to enact a given workflow process. Constructing process models is a complicated and time-consuming task that requires high expertise. Therefore, techniques for discovering process models have been developed. In this paper, we extend the execution of an activity as a time interval, and present a new method for synthesizing process models from sets of systems' audit log. A process model graph (directed graph) generated by the method for a process captures all its executions and dependencies that are present in the audit log, and preserves existing parallel. We compare the model graphs generated by the method (interval) to non-interval methods. The observation is that the graphs are more faithful, the number of excess and missing edges is consistently smaller and it depends on the size and quality of the log. © 2011 Springer-Verlag Berlin Heidelberg.