![](bg1.png)
DataFlow Analysis in BPMN Models
Anass Rachdi, Abdeslam En-Nouaary and Mohamed Dahchour
Institut National des Postes et T
´
el
´
ecommunications, 2, av ALLal EL Fassi Madinat AL Irfane, Rabat, Morocco
Keywords:
BPMN, Business Process Modeling, Formal Verification, Dataflow Anti-patterns, Information Systems.
Abstract:
Business Process Management and Notation (BPMN) is the defacto standard used in enterprises for model-
ing business processes.However, this standard was not provided with a formal semantics, which makes the
possibility of analysis limited to informal approaches such as observation. While most of the existing formal
approaches for BPMN models verification focus on the control-flow, only few has treated the data-flow angle.
The latter is important since the correct execution of activities in BPMN models is based on data’s availability
and correctness. In this paper, we present a new approach that uses the DataRecord concept, adapted for the
BPMN standard. The main advantage of our approach is that it locates the stage where the data flow anomaly
has taken place as well as the source of data flow problem. Therefore the designer can easily correct the data
flow anomaly.The model’s data flow problems are detected using an algorithm specific for the BPMN standard.
1 INTRODUCTION
Business Process Management (BPM), is a manage-
rial approach that enables an organization to ensure
that its processes are implemented effectively and ef-
ficiently. Therefore, it brings an additional value to
organizations by improving their performance, pro-
ductivity and customer services quality. One of the
most important phases that constitutes the life cycle
of Business Process Management is Business Pro-
cess Modeling. The latter is considered essential
for designing and analyzing business process mod-
els that compose information systems. It involves
the use of simple and intuitive modeling languages
that makes models understandable by all business ac-
tors (Business analysts, Technical developers, final
users. . .). One of the latest languages that verifies
these criteria is Business Process Modeling and No-
tation (BPMN 2.0) (OMG, 2011). It is an adopted
standard in both academia and industry that was de-
signed to provide a graphical notation for XML-based
business process languages, like Business Process Ex-
ecution Language (BPEL) (OASIS, 2007). However,
BPMN defines the execution semantics of flow ele-
ments with their data needs and data results only in-
formally, in a textual representation (Stackelberg et
al., 2014), which limits verification to using solely
informal techniques such observation and inspection.
Formal methods help us avoid flow control anomalies
as well as data flow errors. Since several approaches
have addressed the control flow problems (deadlock,
livelock . . .) (Dijkman et al., 2007),(PYH. Wong,
2008),(J.Ye et al., 2008),(Rachdi et al., 2016), we
will focus in this paper on formal methods that deal
with dataflow anomalies (Missing, lost, redundant
and inconsistent data. . .). In order to analyze formally
dataflow in BPMN, we usually define a mapping from
the graphical notation to a formal language such as
Petri Nets (PN). If we adopt the Petri net approach ex-
plained in (Stackelberg et al., 2014), we would have
to go through several and complex steps (Definition
of mapping & Unfolding rules, BPMN to PN trans-
formation, Process-specific anti-patterns generation,
Model checking . . .) before we get the dataflow er-
rors made in the BPMN model. Therefore, we have
taken a different approach that can detect the anti-
patterns representing dataflow anomalies using data
record concept (Kabbaj et al., 2015). The latter is di-
rect and simple and explains to the designer the origin
of the anomaly so he/she can fix it easily in remodel-
ing phase.In addition, this concept does not need sev-
eral operations to complete the desired analysis.
The remainder of this paper is structured as fol-
lows: The next section proposes the work related
to our approach. Section 3 introduces the technical
background needed for the rest of the paper; it is di-
vided into three major parts. The first one presents
BPMN elements as well as their main properties. The
second one presents the dataflow anti-patterns that
have to be avoided during the design phase and the
last one introduces the datarecord concept (Kabbaj et
al., 2015). Section 4 presents our contribution for the
Rachdi, A., En-Nouaary, A. and Dahchour, M.
DataFlow Analysis in BPMN Models.
DOI: 10.5220/0006271202290237
In Proceedings of the 19th International Conference on Enterprise Information Systems (ICEIS 2017) - Volume 2, pages 229-237
ISBN: 978-989-758-248-6
Copyright © 2017 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
229