By-products of Process Mining | BPMN skeleton

The tools that we use for Process Mining in my team, have a powerful configuration system, allowing to fine-tune the activities that the analyst is willing to capture, for each of the defined processes.

A careful selection of the activities is important to limit the volume of data extracted and to avoid a useless clutter of the model.

Among the outputs extracted from the target ERP system (XES and CSV log files), there is a skeletal BPMN definition of the process, created according to the BPMN specifications for the model interchange.

This draft model can be used in two ways:

a) to share with the client an overview of the mining scope, in order to gain approval on the depth of analysis that the extracted data will allow;

b) to check, before running the extraction, if the configuration created is sound.

Here is an example of the draft BPMN model obtained:

As you can see, the model lacks the sequence flows connecting the various tasks. It will be the analyst, after the completion of the mining phase, that will complete the model, perhaps enriching it with some of the findings.

It’s interesting to remind that each activity is defined purely in terms of data change events in the ERP database and that the activity is placed in the proper process context using a series of “back-links” also configured in the extraction phase.