Page 356 -

338 14 Epilogue

However, most data mining techniques are not process-centric. Fortunately, process
mining provides a link between both disciplines. Like other BPM approaches, pro-
cess mining is process-centric. However, unlike most BPM approaches, it is driven
by factual event data rather than hand-made models. Hence, process mining can be
seen as a bridge between the preliminaries presented in Chaps. 2 and 3.
In Part II, we focused on the most challenging process mining task: process dis-
covery. First, we discussed the input needed for process mining (Chap. 4). Then, we
presented a very basic algorithm (Chap. 5) followed by an overview of more pow-
erful process discovery techniques (Chap. 6). Unlike basic data mining techniques
such as decision tree and association rule learning, process discovery problems are
characterized by a complex search space as is illustrated by the many workflow pat-
terns. Whereas the aim of many data mining techniques is to be able to deal with
many records or many variables, the main challenge of process discovery is to ade-
quately capture behavioral aspects.
Process mining is not limited to process discovery. In fact, process discovery is
just one of many process mining tasks. Therefore, Part III expanded the scope of
process mining into several directions. These expansions have in common that the
event log and the process model are tightly coupled, thus allowing for new forms
of analysis and support. Chapter 7 presented various conformance checking tech-
niques. As shown in Chap. 8, the organizational perspective, the case perspective,
and the time perspective can be added to discovered process models or used to create
complementary models. Recommendations and predictions (based on a combination
of historic event data and partial traces of running cases) are examples of the oper-
ational support functionalities described in Chap. 9. Chapters 7, 8, and 9 illustrate
the breadth of the process mining spectrum.
Part IV aimed to provide useful hints when applying process mining in practice.
Chapter 10 discussed tool support for process mining. In Chaps. 11 and 12,we
described two types of processes (“Lasagna processes” and “Spaghetti processes”)
that need to be handled differently.
In this last part (Part V), we take a step back and reflect on the material presented
in the preceding parts. For example, Chap. 13 compared business process models,
business process analysis, and business process support with geographic maps and
navigation systems. This comparison revealed limitations of current BPM practices
and confirmed the potential of process mining to “breathe life” into process models.
Process mining provides not only a bridge between data mining and BPM; it also
helps to address the classical divide between “business” and “IT”. IT people tend
to have a technology-oriented focus with little consideration for the actual business
processes that need to be supported. People focusing on the “business-side” of BPM
are typically not interested in technological advances and the precise functionality of
information systems. The empirical nature of process mining can bring both groups
of people together. Evidence-based BPM based on process mining helps to create a
common ground for business process improvement and information systems devel-
opment.