=Paper=
{{Paper
|id=Vol-1848/CAiSE2017_DC_Paper2
|storemode=property
|title=Towards the Design of a Process Mining-Enabled Decision Support System for Business Process Transformation
|pdfUrl=https://ceur-ws.org/Vol-1848/CAiSE2017_DC_Paper2.pdf
|volume=Vol-1848
|authors=Christian Fleig
|dblpUrl=https://dblp.org/rec/conf/caise/Fleig17
}}
==Towards the Design of a Process Mining-Enabled Decision Support System for Business Process Transformation==
https://ceur-ws.org/Vol-1848/CAiSE2017_DC_Paper2.pdf
Towards the Design of a Process Mining-Enabled
Decision Support System for Business Process
Transformation
Christian Fleig
Supervisor: Alexander Maedche
Karlsruhe Institute of Technology, Institute of Information Systems and Marketing (IISM),
Information Systems and Service Design (ISSD), Karlsruhe, Germany
christian.fleig@kit.edu
Abstract. Current approaches to business process transformation rely on nor-
mative “de jure” process models which are derived manually, costly to create,
error-prone, idealistic, and often deviating from process reality. Theoretically
grounded in organizational contingency theory, decision support systems
(DSSs) and business process management literature, this design science project
proposes the development of a DSS which incorporates bottom-up process min-
ing in addition to other top-down sources of process knowledge for business
process transformation. The DSS is intended to provide support in both the se-
lection of an appropriate target process design as well as transformation support
on the task level. This design project is conducted within a large-scale digital
transformation project of a leading German manufacturing corporation to rede-
sign business processes including the migration of the current SAP R/3 system
to the SAP S/4 HANA business suite. Therefore, this design science research
(DSR) utilizes a real-life event log comprising transaction data from multiple
ERP systems as data source for the process mining module of the DSS and the
later evaluation of the artifact. This proposal motivates both the practical and
theoretical need for process mining-enabled decision support in business pro-
cess transformation. Further, this proposal highlights research gaps, outline the
DSR approach, and introduces the meta-requirements and the technical concep-
tualization of the DSS.
Keywords: business process management, process mining, decision support
system, design science, process transformation, contingency theory
1 Motivation
Organizational success in the 21st century vitally depends on the ability to navigate
in increasingly dynamic environments by constantly adapting the organizational de-
sign. Organizations perform business process transformation to restore the fit between
the environment and the organizational design to remain competitive (e.g., [1–3]).
Numerous environmental shifts and technological innovations fundamentally change
organizational boundary conditions (e.g., [4–6]), and thus require organizations to
X. Franch, J. Ralyté, R. Matulevičius, C. Salinesi, and R. Wieringa (Eds.):
CAiSE 2017 Forum and Doctoral Consortium Papers, pp. 170-178, 2017.
Copyright 2017 for this paper by its authors. Copying permitted for private and academic purposes.
transform business processes in response to these internal and external changes. For
instance, digital transformation imposes an “inescapable” discontinuity (e.g., [7]), and
organizations engage in digital transformation with the replacement of physical pro-
cesses by information-based processes [8] to enrich purely physical products with
value-creating services in response to the emergence of technological innovations as
well as changed market economies (e.g., [2]).
However, the transformation of business processes from a current process design X
to a new target process design X’ requires a comprehensive understanding of the real-
world execution of current processes [9] in order to make solid transformation deci-
sions. Furthermore, a clear decision support on how to reach the target design X’ is
required. Nevertheless, organizations frequently do not to meet these prerequisites for
business process transformation. Many organizations possess only limited process
insights and a narrow understanding of process execution paths [10–12]. In particular,
the understanding of processes in transformation endeavors often relies on feeble “de
jure” [13] process models. These “de jure” models are normative and prescriptive in
nature, and purely describe how processes were originally intended to be executed
during design-time. “De jure” documentations usually merely contain ideal process
executions, while most process variants and deviations from ideal to-be specifications
are ignored [14]. Therefore, van der Aalst [13] finds the currently prevailing ap-
proaches to process modeling to be “disconnected” from process realities. Further-
more, and in addition to weaknesses concerning content and completeness, “de jure”
models are derived manually and top-down in a time-consuming and error-prone doc-
umentation procedure by process stakeholders, which introduces another dimension
of insufficiency in terms of costs. In sum, “de jure” process models provide an unsuit-
able foundation for decision-making in process transformation.
2 Proposal and Research Question
The design science approach in this project tries to provide a solution to these limi-
tations to organizational transformation capabilities. A possible way to overcome the
various insufficiencies of “de jure” process models is to exploit the increasing availa-
bility of process data from numerous sources internal and external to organizations
[5]. These different data pools which store process information such as information
systems (IS) might be utilized to enrich the traditional top-down “de jure” models
with bottom-up “de facto” process information in decision-making. The core idea of
this research is therefore to explore how “de jure” top-down process information
might be enriched by data-driven, bottom-up and “de facto” process information for
decision-making in business process transformation.
To operationalize this idea, process mining [15–17] provides a technology, which
aims at the automatic discovery of processes from event logs stored in organizational
IS [17]. Therefore, process mining yields descriptive and positive “de facto” process
analyses which precisely and thoroughly capture process realities [14].
171
Thus, the aim of this design science research (DSR) project is to provide organiza-
tions with a process mining-enabled decision support system for business process
transformation. The overarching research question of this DSR project becomes:
Research Question: How to design a process mining-enabled decision support sys-
tem to support organizations in transformation of business processes?
Process mining has reached a state of maturity by providing widespread tools and
techniques to model, monitor, and improve organizational processes, and delivers
mature techniques and algorithms to turn data into process knowledge [4]. Current
research on process mining has been rather abundant in exploring techniques and
algorithms to process event logs (e.g., [18, 19]), in developing software applications
and tools (e.g., [9, 16]), in identifying challenges when turning data into process in-
formation (e.g., [20]), and in creating visual process representations (e.g., [16, 21]).
However, these contributions focus on the “pre-processing” and the mining phase of
process mining, while the “interpretation phase” [22] or post-mining phase received
significantly less scholastic attention. Thus, further research is needed on the “post-
mining phase” to provide an answer to the question of how organizations can actually
employ process mining results in business process transformation.
3 Research Methodology
The proposed DSS will be developed in a design science approach. The seminal
works by Hevner [23] and Kuechler and Vaishnavi [24] suggest to perform DSR pro-
jects in cycles in “build-and-evaluate loops” [23] to iteratively arrive at an optimized
artifact instantiation. Thus, this research will conduct two sequential design cycles.
The first initial design cycle comprises a phase to create problem awareness, the
formulation of a suggested problem solution, the development of a first DSS proto-
type, as well as an evaluation phase to discover potential for improvement of the pro-
totype [25]. In the problem awareness phase of the first cycle, an industry alliance
was formed. Further, a structured literature review on the search string “process min-
ing” was conducted to validate this proposal as a previously unexplored research gap.
In the proposal phase, a further unstructured literature review was conducted to derive
preliminary meta-requirements (MRs) for the DSS instantiation. As a next step in the
development phase of cycle one, a preliminary DSS prototype will be developed for
an order-to-purchase process in the underlying SAP R/3 system. Finally, the evalua-
tion phase will follow the design science evaluation framework as provided by Vena-
ble et al. [25] and evaluate the DSS via interviews with process managers and process
experts for different processes at different companies of the research site. The proto-
type will therefore be evaluated in the planning of several real-life transformations of
business processes for different companies. The authors in Venable et al. [25] require
an evaluation to demonstrate the artifact ability to achieve the intended purpose as
well as evidence of problem resolution through the artifact in terms of utility, quality,
and efficacy. Furthermore, Venable et al. [25] require developed artifacts to be com-
pared against similar and previously existing solutions. Thus, the evaluation of the
172
DSS will be conducted to judge the superiority of the DSS to current approaches to
business process transformation without the decision support given by the process
mining-enabled artifact to evaluate utility and efficacy in business process transfor-
mation. In order to quantify these constructs, interviews with an expert group of
around 8-10 process managers in the project team and selected individuals from a
pool of about 80 specialized process experts across several companies will be con-
ducted to evaluate the helpfulness of the DSS in business process transformation.
Besides, the quality of the prototype and the final software artifact will be evaluated
according to the properties of “quality” in the ISO25010:2011 standard plus addition-
al quality dimensions from a literature review such as usability, user satisfaction and
perceived usefulness.
The second design cycle will further enhance the DSS artifact instantiation by
building on the results from the previous design cycle. In addition to a further refine-
ment of problem awareness, the proposal for a solution and the development, cycle
two concludes with the finalization of a DSS software artifact and the subsequent
summative evaluation [25].
4 Data
An industry cooperation with the IT service company of a large German manufactur-
ing corporation was formed to gain access to several manufacturing companies as
research sites to gather qualitative and quantitative data throughout the phases of the
DSR project. In 2014, the corporation consisted of several sub-companies operating
globally with more than 8.400 employees and about 1bn Euro in turnover.
The industry partner provided an event log of several hundreds of gigabytes of raw
data from different SAP R/3 ERP systems for the process mining module of the DSS,
plus the additional possibility to conduct qualitative first-hand research such as inter-
views and quantitative surveys to later evaluate the DSS artifact in a real-life context.
This DSR project is conducted within a large-scale digital transformation project to
redesign business processes across sub-companies. The digitalization project further
comprises the migration from the status-quo SAP R/3 ERP to the future SAP S/4
HANA architecture. To the best of knowledge, this is the first process mining and
decision support research project being able to conduct studies in such a context. As a
further research gap in process mining literature, the more advanced process mining
techniques have not been tested sufficiently on real-life process data [16] due to a lack
of event logs available to researchers. This contribution therefore uses the real-life
event log retrieved from the industry partnership, and thus overcomes the weaknesses
of many process mining contributions when relying on synthetic, simulated data.
5 Preliminary Results: Meta-Requirements for the DSS
The following section describes the set of meta-requirements which have been identi-
fied in the first design cycle. Figure 1 illustrates the conceptualization of the meta-
requirements.
173
Process
T ransformation:
X
t o
X‘
Process
E vent
L og
Process
Definitions,
Process
M ining Documentations,
Human
Process
Knowledge
Process
M odels
and
Process
M etrics
MR3:
Combination
of
bottom-‐up
and
t op-‐down
process
information
MR4:
Process
T ype MR8:
Goal
Prioritization
Process
Mining-‐
MR5:
Contextual
Factors
/
Process
Enabled
Decision
MR9:
Matching
L ogic
Characteristics
Support
System
for
MR7:
Transformation
G oals
Business
Process
DP(7,1):
Organization-‐Level
Goals
Transformation
DP(7,2):
Process-‐Level
Goals MR5:
Process
R epository
Fig. 1. Meta-Requirements of the Decision Support System
Organizational process knowledge might either be prescriptive (top-down) or de-
scriptive (bottom-up), and might be dispersed across different tangible or intangible
“storage locations” in the organization. Therefore, the DSS needs to be able to re-
trieve process knowledge from different sources, and to combine the different types of
process information before deriving transformation decisions.
A potential source of process information is bottom-up process knowledge stored
in IS such as ERP systems. These sources include data generated by systems during
process execution, such as event log tables within the ERP systems.
MR1: The DSS needs to incorporate descriptive bottom-up process information
These quantitative sources capture process executions “as-is”. An exclusive reli-
ance upon process mining in decision-making for business process transformation
yields merely an incomplete picture of process realities. Process mining captures only
information on process activities within the information system (e.g., [17]), and event
logs merely contain a subset of all possible process facets [14, 17]. Insights gained by
process mining might be incomplete due to shadow process steps which are not rec-
orded in the system event log. Hence, additional process knowledge needs to be re-
trieved from top-down sources including qualitative process documentations, defini-
tions, and intangible human process knowledge, leading to MR2:
MR2: The DSS needs to incorporate prescriptive top-down process information
Furthermore, both forms of process knowledge need to complement each other to
overcome the mutual insufficiencies. Hence, as bottom-up (descriptive) in MR1 and
top-down (prescriptive) process information in MR2 each deliver an incomplete anal-
174
ysis of processes in isolation, both sources need to be combined by the DSS in deci-
sion-making to complement each other.
MR3: The DSS needs combine descriptive bottom-up and prescriptive top-down
process knowledge
Besides, as this research is embedded in organizational contingency theory [26] by
Donaldson [26], which ultimately requires activities of business process management
to consider the respective circumstances and contexts of processes, the DSS needs to
include the diversity and different types of processes in organizations into decision-
making. Therefore, the DSS needs to incorporate process information and to capture
information on process type, key process dimensions, and characteristics to provide
decision support depending on the respective type of process which is to be trans-
formed.
MR4: The DSS needs to incorporate the type of process into decision-making
Further required by the theoretical embedding into organizational contingency the-
ory [26], process contexts need to be taken into account. The work by vom Brocke et
al. [27] requires BPM to be contextual to be effective and defines context as the en-
tirety of factors concerning goals, processes, the organization itself, as well as the
environment surrounding the organization.
MR5: The DSS needs to incorporate process context factors and process charac-
teristics into decision-making
In addition to the status quo-oriented meta-requirements, an additional meta-
requirement is established concerning the future process state. The DSS needs to pos-
sess a repository of potential standard specifications concerning the future target pro-
cess design from which an optimal process design in X’ can be chosen.
MR6: The DSS needs a repository of process designs for the future design in X’
However, to select an “optimum” process design in X’ from the repository, needs
to be given input concerning transformation goals.
MR7: The DSS needs information concerning transformation goals
Most approaches in BPM usually involve strategic process goals [1, 28] which are
compatible with the overall organization strategy [29]. Design principle 𝐷𝑃!,! conse-
quently demands the DSS to include strategic organizational goals into decision-
making. Furthermore, 𝐷𝑃!,! imposes another requirement in the form of process-level
transformation goal input.
𝐷𝑃!,! : The DSS incorporates strategic organizational-level goals
𝐷𝑃!,! : The DSS incorporates process-level transformation goals
A challenge with transformation goals however is their mutual incompatibility as
well as different levels of importance allocated to the goals by the DSS users. Deci-
sion-making concerning process goals requires multiple criteria decision-making. In
turn, this requires the DSS to weigh goals according to importance in advance [30] to
give one goal priority over another via an importance ranking (prioritization) among
these goals [1].
MR8: The DSS needs an importance ranking among goals to select among target
process design alternatives
Finally, the DSS additionally needs to incorporate a matching logic to determine
the most suitable target process design X’ in the repository. The matching logics
175
needs to compare the status-quo process model X created by the combination (MR3)
of bottom-up (MR1) and top-down (MR2) process knowledge with the process mod-
els stored in the repository (MR4) and select an appropriate target model X’.
MR9: The DSS needs matching logic to select an appropriate process design X’
Figure 2 provides an overview of the technical conceptualization of the DSS under
consideration of the derived meta-requirements in the proposal phase of cycle one.
Decision
Support
Module
DSS
Output:
Selection
of
Appropriate
X ‘
and
Transformation
Support
on
Task-‐Level
Process
M atching
Module
Process
M atching
Algorithm
Process
Combination
Module Process
Repository
Module
Process
M ining
Module Combined
Model
( X)
(BPMN) MR6:
Future
Process
Design
MR1:
De-‐Facto
Process
(MR3) Target
Repository
Knowledge,
Bottom-‐Up (X‘)
(BPMN)
Process
M odel
(BPMN) User
Interface
Module
Process
Visualization
Additional
Top-‐Down
Process
Knowledge
( MR2)
Event
Log
Process
T ype
(MR4)
Merging
Logics
Contextual
Factors
&
Process
Data
Tables Characteristics
(MR5)
Data
Extraction
Algorithm Goals
( MR7)
Information
Information
Systems
(ERP)
Systems
Layer
Fig. 2. Technical Conceptualization of the DSS based on Meta-Requirements
6 Conclusion
The ubiquitous need for organizational adaptation in response to changes in the trinity
of business, technology, and humans within and around organizations gives a power-
ful impetus to fundamentally transform organizational business processes. The DSS
proposed by this project is created to improve the current way organizations approach
business process transformation by providing support in the selection of an appropri-
ate target process model as well as transformation support on the task level. By incor-
porating “de facto”, bottom-up models derived in process mining instead of purely
relying on “de jure” documentations in decision-making, the DSS might considerably
improve the organizational capability to transform processes. Among the future chal-
lenges is the specification of how the task-level transformation support might be im-
plemented.
176
References
[1] P. Afflerbach and F. Leonhard, “Customer Experience Versus Pro-
cess Efficiency: Towards an Analytical Framework About Ambidextrous BPM:
Completed Research Paper,” International Conference on Information Systems,
2016.
[2] E. Piccinini, R. W. Gregory, and L. M. Kolbe, “Changes in the Producer-
Consumer Relationship - Towards Digital Transformation,”
Wirtschaftsinformatik Proceedings, vol. 2015, no. 109, 2015.
[3] H. C. Lucas Jr., R. Agarwal, E. K. Clemons, O. A. El Sawy, and B. Weber,
“Impactful Research on Transformational Information Technology: An Oppor-
tunity to Inform New Audiences,” Management Information Systems Quarterly,
vol. 37, no. 2, pp. 371–382, 2013.
[4] A. Braganza, L. Brooks, D. Nepelski, M. Ali, and R. Moro, “Resource man-
agement in big data initiatives: Processes and dynamic capabilities,” Journal of
Business Research, vol. 70, pp. 328–337, 2017.
[5] C. Loebbecke and A. Picot, “Reflections on societal and business model trans-
formation arising from digitization and big data analytics: A research agenda,”
The Journal of Strategic Information Systems, vol. 24, no. 3, pp. 149–157, 2015.
[6] G. Seeger and M. Bick, “Mega and Consumer Trends - Towards Car-
independent Mobile Applications,” International Conference on Mobile Busi-
ness, p. 27, 2013.
[7] E. Henriette, M. Feki, and I. Boughzala, “The Shape of Digital Transformation:
A Systematic Literature Review,” Mediterranean Conference on Information
Systems, 2015.
[8] D. W. Straub and R. T. Watson, “Research Commentary: Transformational
Issues in Researching IS and Net-Enabled Organizations,” Information Systems
Research, vol. 12, no. 4, pp. 337–345, 2001.
[9] A. Tiwari, C. J. Turner, and B. Majeed, “A review of business process mining:
State of the art and future trends,” Business Process Management Journal,
vol. 14, no. 1, pp. 5–22, 2008.
[10] R. Gopal, J. R. Marsden, and J. Vanthienen, “Information mining — Reflections
on recent advancements and the road ahead in data, text, and media mining,”
Recent Advances in Data, Text, and Media Mining & Information Issues in Sup-
ply Chain and in Service System Design, vol. 51, no. 4, pp. 727–731, 2011.
[11] F. Caron, J. Vanthienen, and B. Baesens, “Comprehensive rule-based compli-
ance checking and risk management with process mining,” Decision Support
Systems, vol. 54, no. 3, pp. 1357–1369, 2013.
[12] W. van der Aalst and A. Weijters, “Process mining: a research agenda,” Process
/ Workflow Mining, vol. 53, no. 3, pp. 231–244, 2004.
[13] W. van der Aalst, “"Mine your own business": using process mining to turn big
data into real value,” Proceedings of the 21st European Conference on Infor-
mation Systems, http://wwwis.win.tue.nl/~wvdaalst/publications/p747.pdf, 2013.
177
[14] W. van der Aalst, “Process Mining in the Large: A Tutorial,” in Lecture notes in
business information processing, Business Intelligence, E. Zimányi, Ed., Cham:
Springer International Publishing, 2014, pp. 33–76.
[15] S. Schönig, C. Cabanillas, S. Jablonski, and J. Mendling, “A framework for
efficiently mining the organisational perspective of business processes,” Deci-
sion Support Systems, vol. 89, pp. 87–97, 2016.
[16] W. van der Aalst et al., “Business process mining: An industrial application,”
Information Systems, vol. 32, no. 5, pp. 713–732, 2007.
[17] W. van der Aalst, Process mining: Discovery, conformance and enhancement of
business processes. Berlin, Heidelberg, New York: Springer, 2011.
[18] M. de Leoni, F. M. Maggi, and W. M. van der Aalst, “An alignment-based
framework to check the conformance of declarative process models and to pre-
process event-log data,” Information Systems, vol. 47, pp. 258–277, 2015.
[19] J. de Weerdt, M. de Backer, J. Vanthienen, and B. Baesens, “A multi-
dimensional quality assessment of state-of-the-art process discovery algorithms
using real-life event logs,” Information Systems, vol. 37, no. 7, pp. 654–676,
2012.
[20] S. Suriadi, R. Andrews, A. ter Hofstede, and M. T. Wynn, “Event log imperfec-
tion patterns for process mining: Towards a systematic approach to cleaning
event logs,” Information Systems, vol. 64, pp. 132–150, 2017.
[21] W. van der Aalst, T. Weijters, and L. Maruster, “Workflow mining: Discovering
process models from event logs,” IEEE Trans. Knowl. Data Eng., vol. 16, no. 9,
pp. 1128–1142, 2004.
[22] R. Berlanga and V. Nebot, Data Mining: Concepts, methodologies, tools, and
applications. XML Mining for Semantic Web. Hershey, PA: IGI Global, 2012.
[23] A. R. Hevner, S. T. March, J. Park, and S. Ram, “Design Science in Information
Systems Research,” Management Information Systems Quarterly, vol. 28, no. 1,
pp. 75–105, 2004.
[24] B. Kuechler and V. Vaishnavi, “On theory development in design science re-
search: anatomy of a research project,” European Journal of Information Sys-
tems, vol. 17, no. 5, pp. 489–504, 2008.
[25] J. Venable, J. Pries-Heje, and R. Baskerville, “FEDS: A Framework for Evalua-
tion in Design Science Research,” European Journal of Information Systems,
vol. 25, no. 1, pp. 77–89, 2016.
[26] L. Donaldson, The contingency theory of organizations: Sage, 2001.
[27] J. Vom Brocke, S. Zelt, and T. Schmiedel, “On the role of context in business
process management,” International Journal of Information Management, vol.
36, no. 3, pp. 486–495, 2016.
[28] S. Limam Mansar, H. A. Reijers, and F. Ounnar, “Development of a decision-
making strategy to improve the efficiency of BPR,” Expert Systems with Appli-
cations, vol. 36, no. 2, pp. 3248–3262, 2009.
[29] J. Vom Brocke et al., “Ten principles of good business process management,”
Business Process Management Journal, vol. 20, no. 4, pp. 530–548, 2014.
[30] J. Pries-Heje and R. Baskerville, “The design theory nexus,” Management In-
formation Systems Quarterly, pp. 731–755, 2008.
178