Extracting Connection Types in Process Models Discovered by Using From-to Chart Based Approach - Developing Concepts in Applied Intelligence - page 64

In [9] , Weijters and Aalst propose a rediscovery technique to deal with noise and to
validate workflow processes by uncovering and measuring the discrepancies between
the prescriptive models and actual process executions. Detecting concurrency appears
as one of the fundamental concerns. Therefore AND/OR connectors are aimed to be
explicitly distinguished in the process model. Moreover, local and global metrics are
proposed to find explicit representations for a broad range of process models. These
metrics are revised in [1] . In [10] , the proposed method, named alpha algorithm, seeks
to rediscover sound Workflow nets. Given such an algorithm, it is aimed to find the
class of workflow nets which can be rediscovered. As a way of representation, it
generates concrete Petri nets for a broad range of process models rather than a set of
dependency relation between events like in [9] .
As a GA implementation in facility layout problem (FLP), Sarker et al. attempts to
handle locating multiple identical machines in a linear layout, which is called tertiary
assignment problem (TAP) in [12] . It is attempted to solve this problem type by
identifying sets of identical machines which may be partitioned into individual,
unique machines. In order to relax the underlying FLP and provide approximate
solutions, amoebic matrix is used.
5 Conclusion
Process design is a complicated, time-consuming phase and typically, there may arise
incompatibilities between process design and process enactment. Process mining
techniques aim to handle such incompatibilities by extracting information on the
process from the event logs. In this work, we present an enhancement on a hybrid
process mining approach that uses from-to chart, in order to extract connection types
and to form a process model. From-to chart, which is a fundamental analytical tool
used in tracing material handling routes at production floor, is adapted to process
modeling domain in order to analyze transitions among activities. The activity
sequence, which is rearranged in a straight-line form by developed GA engine,
constitutes the backbone of dependency/frequency graph without compromising
accuracy. In order to construct the process model, the generated
dependency/frequency graph is converted into control flow graph according to AND
threshold. The applicability of the approach is demonstrated on a case study.
References
[1] Măruşter, L., Weijters, A.J.M.M.T., van der Aalst, W.M.P., van den Bosch, A.: Process
mining: Discovering direct successors in process logs. In: Lange, S., Satoh, K., Smith,
C.H. (eds.) DS 2002. LNCS, vol. 2534, pp. 364-373. Springer, Heidelberg (2002)
[2] Esgin, E., Senkul, P.: Hybrid Approach to Process Mining: Finding Immediate Successors
of a Process by Using From-to Chart. In: Int. Conf. on Machine Learning and
Applications, pp. 664-668. IEEE Computer Society, Los Alamitos (2009)
[3] Esgin, E., Senkul, P., Cimenbicer, C.: A hybrid approach for process mining: Using from-
to chart arranged by genetic algorithms. In: Graña Romay, M., Corchado, E., Garcia
Sebastian, M.T. (eds.) HAIS 2010. LNCS, vol. 6076, pp. 178-186. Springer, Heidelberg
(2010)