Page 284 -

266 10 Tool Support

Fig. 10.3 Screenshot of ProM 5.2 showing two of the 286 plug-ins. The bottom window shows the
conformance checker plug-in while checking the fitness of event log L full described in Table 7.1
and WF-net N 2 depicted in Fig. 7.2. The plug-in identifies the conformance problem (the log and
model disagree on the position of d) and returns a fitness value computed using the approach
presented in Sect. 7.2: fitness(L full ,N 2 ) = 0.95039195. The left window shows the trace clustering
plug-in using Self Organizing Maps (SOM) to find homogeneous groups of cases. The largest
cluster contains 641 cases. These are the cases that were rejected without a thorough examination
(i.e., traces σ 1 , σ 3 , σ 13 in Table 7.1)

discovery algorithms presented in Chap. 6 (genetic mining, heuristic mining, fuzzy
mining, etc.) corresponds to one of the 47 mining plug-in of ProM 5.2. The replay
approach presented in Sect. 7.2 describes only one of the conformance checking
techniques supported by ProM’s conformance checker plug-in [80]. This illustrates
that this book can only cover a fraction of the functionality provided by ProM.
The spectacular growth of the number of plug-ins in the period from 2004 to
2009 illustrates that ProM realized its initial goal to provide a platform for the de-
velopment of new process mining techniques. ProM has become the de facto stan-
dard for process mining. Research groups from all over the globe contributed to
the development of ProM and thousands of organizations downloaded ProM. In the
same period we applied ProM in more than 100 organizations, e.g., in the context of
joint research projects, Master projects, and consultancy projects. The large number
of plug-ins and the many practical applications also revealed some problems. For
example, ProM 5.2 can be quite confusing for the inexperienced user who is con-
fronted with almost 300 plug-ins. Moreover, in ProM 5.2 (and earlier versions) the
user interface and the underlying analysis techniques are tightly coupled, i.e., most
plug-ins require user interaction. To be able to run ProM remotely and to embed pro-
cess mining functionality in other systems, we decided to completely re-implement
ProM from scratch. This allowed us to learn from earlier experiences and to develop
a completely new architecture based an improved plug-in infrastructure.