5th SIKS/BENAIS Conference on Enterprise Information Systems




       Towards Robust Conformance Checking
               (Extended Abstract)

         A. Adriansyah, B.F. van Dongen, and W.M.P. van der Aalst

                Department of Mathematics and Computer Science
                       Eindhoven University of Technology
               P.O. Box 513, 5600 MB Eindhoven, The Netherlands
              a.adriansyah,b.f.v.dongen,w.m.p.v.d.aalst@tue.nl

    The growing complexity of processes in many organizations stimulates the
adoption of business process management (BPM) techniques. Process models
typically lie at the basis of these techniques and generally, the assumption is
made that the operational business processes as they are taking place in practice
conform to these models. However, recent experience has shown that this often
isn’t the case. Therefore, the problem of checking to what extent the operational
process conforms to the process model is increasingly important.
    In [1], we provide a robust method for calculating conformance between a
log and a process model. First, we introduce flexible models that provide an
abstraction of many languages and allow for the modeling of complex control
flow constructs, such as OR-split/joins and multiple tasks that represent the
same activity. We provide semantics for these models, but without specifying
how to execute them. Instead, we show that in the context of a case that has
been recorded in the log, we can construct instances of the model that maxi-
mize certain conformance metrics. Finally, using experiments on simulated data
(comparable in size to real-life data sets), we show that our approach calculates
fitness correctly in the presence of complex constructs, where existing approaches
do not.
    Given a process model and an event log, our approach does not only solve the
problem of having inaccurate conformance results, but also opens possibilities
to compare conformance values for a given log between different models made in
different languages. With existing approaches, conformance comparison between
two process models and a given log require both models to be created in the same
process modeling language. We soften the requirement such that models can be
in different modeling languages as long as they can be represented as abstract
models.
    The work presented in this paper provides a solid basis for robust confor-
mance checking. Since our flexible models do not have executable semantics,
we do not rely on state-space exploration (which is required in Petri-net based
conformance checking). Using the approach presented in [1], we plan to define
metrics that capture different dimensions of conformance. Furthermore, our ap-
proach can easily be extended to identify the “skipping” of activities, i.e. by
identifying which tasks were executed but not logged.
    Our approach is fully implemented in Version 6 of the ProM framework, that
can be obtained from http://www.processmining.org, evaluated using simulated




                                      111
Proceedings




              event logs and compared against an existing conformance technique based on
              Petri nets. By experiments, it is proven that the proposed approach provide
              better insights to conformance between a given process model and an event log,
              particularly when advanced control flow constructions (e.g. OR split, OR join)
              exist in the model.


              References
              1. A. Adriansyah, B.F. van Dongen, and W.M.P. van der Aalst. Towards Robust Con-
                 formance Checking. In Proceedings of the 6th International Workshop on Business
                 Process Intelligence (BPI 2010 - accepted), 2010.




                                                    112