Temporal analysis of process logs: A case study (Extended Abstract)

In the last years there has been increasing interest to the analysis of process logs and several techniques such as workflow mining have been proposed aimed at automatically deriving the underlying workflow models. However, current approaches essentially disregard an important information contained in these traces, the timestamps, used only to define a sequential ordering of the performed tasks. In this work we try to overcome these limitations, by explicitly including the temporal information in the analysis, that allows to distinguish among different temporal behaviours. That makes it possible to discern between apparently identical process executions that are performed with different transition times between consecutive tasks. Detecting such differences would enable a more accurate comparison between the original workflow design and its actual usage. This work faces the above problem by applying a novel mining paradigm named Time-Annotated Sequences (TAS) aimed at extracting sequential patterns where each transition between two events is annotated with a typical transition time that emerges from input data. We report a real-world case study, in which the TAS mining paradigm is applied to databases of log traces obtained from the actual execution of processes. We believe that this case study not only shows the interestingness of extracting TAS patterns in the workflow context, but, more ambitiously, it opens the way for a novel approach to workflow mining.