=Paper=
{{Paper
|id=Vol-2339/paper2
|storemode=property
|title=Towards a Comprehensive Methodology for Process Mining (short paper)
|pdfUrl=https://ceur-ws.org/Vol-2339/paper2.pdf
|volume=Vol-2339
|authors=Kiarash Diba
|dblpUrl=https://dblp.org/rec/conf/zeus/Diba19
}}
==Towards a Comprehensive Methodology for Process Mining (short paper)==
https://ceur-ws.org/Vol-2339/paper2.pdf
Towards a Comprehensive Methodology for
Process Mining
Kiarash Diba
Hasso Plattner Institute, University of Potsdam, Potsdam, Germany
{kiarash.diba}@hpi.de
Abstract. Process mining exploits data recorded in information sys-
tems of organizations to unleash insight and knowledge into their oper-
ational processes. As process mining techniques are reaching maturity,
their applications are becoming more widespread across various domains.
Therefore, more research on the methodological and practical perspec-
tive is required to steer and guide these applications to successful result.
This position paper sketches the first steps to be taken toward a standard
methodology for process mining.
Keywords: Process Mining · Process Mining Methodology · Process
Mining Reference Model
1 Introduction and Motivation
Process Mining has evolved into a well-known technology to provide valuable
insight into the underlying processes and workflows of organizations. During the
recent years, many process mining techniques and algorithms have been devel-
oped and are reaching maturity and their applications have been investigated
and proved valuable across variety of domains. Despite this level of maturity
in techniques and algorithms, the broader process mining discipline has not yet
matured. Although process mining projects involve various steps and activities
from extracting and preparing required data to providing useful knowledge and
insight, the entire spectrum of activities have not been thoroughly investigated.
Instead most of the academic focus has been concentrated on the development
and improvement of techniques and algorithms. Besides, most case studies and
projects have been carried out in an unstructured and ad-hoc manner involving
a great amount of manual and time intensive work and there is little guidance
on conducting such projects successfully in both industrial and academic set-
tings. Therefore, this work focuses on methodological aspects of process mining
rather than on specific methods, in order to provide well-defined foundations for
process mining. This not only helps practitioners and academics in conducting
successful process mining projects, but positions process mining techniques into
a broader spectrum of process-related knowledge discovery and sheds more light
on steps that have received less research attention. Thus, inspired by related
works in the field of data mining and practical experience, this work takes the
initial steps towards a comprehensive standard methodology for process mining.
S. Kolb, C. Sturm (Eds.): 11th ZEUS Workshop, ZEUS 2019, Bayreuth, Germany, 14-15
February 2019, published at http://ceur-ws.org/Vol-2339
10 Kiarash Diba
The remainder of this paper is structured as follows. The next section dis-
cusses related works and their limitations. Section 3 outlines the approach to be
followed and steps to be taken for establishment of the comprehensive methodol-
ogy, followed by a high-level overview of the initial developments of the method-
ology in section 4. Finally, section 5 concludes the paper.
2 Related work
Methodology provides the theoretical foundation for understanding which meth-
ods, set of methods, or best practices can be applied to a specific case, which is
employed for the design, planning, implementation and achievement of project
objectives [5]. Currently there are few works focusing on methodologies for pro-
cess mining namely the L* lifecycle model [1], Process Diagnostic Method (PDM)
[3], its extension in healthcare domain [6], and PM2 [4]. However, these method-
ologies are not comprehensive and suitable for every project and have a num-
ber of limitations. Besides, they have not been widely evaluated and applied
in various projects. PDM has a narrow scope focusing on a limited number of
capabilities of process mining. Besides, it neglects the importance of business
considerations, planning and domain knowledge [4]. L* lifecycle model primarily
focuses on discovery of a single process model enriched with performance and re-
source information and therefore, it is more suitable for structured processes and
narrowly scoped projects [4]. In addition, non of the two offer sufficient flexibility
and iterations. The sequence of activities suggested are assumed to be followed
rather strictly for every project which is rarely the case in complex projects.
In different projects depending on different requirements, some steps might be
skipped or performed in a different sequence. Although PM2 addresses a few
limitations of the previously mentioned methodologies, it can still be improved
and extended with more flexibility, more detailed and specific steps, techniques,
best practises and practical guidelines. Successful examples of methodologies can
be found in the field of data mining where similar works such as CRISP DM [7]
have been applied successfully for many years in variety of settings.
3 Proposed Approach
In order to construct a comprehensive methodology we will first define method-
ology and clarify what a methodology is and what it should contain and motivate
the use and benefits of such methodologies. Then we will formally compare re-
lated work from both process mining literature and related fields such as data
mining based on their structure, applicability and reputation. We will also ana-
lyze case studies and use cases from a structural point of view before establishing
the methodology. In addition, we will make use of questionnaires among process
mining experts both in academia and industry to consolidate the motivation and
formation of the methodology. Afterwards, the methodology needs to be tested,
evaluated and consequently adjusted followed by continuous refinements. The
Towards a Comprehensive Methodology for Process Mining 11
next section provides an initial high-level overview of the methodology to be
developed.
4 Outline of the methodology
A methodology should be able to be applied to specific projects in different con-
text with different goals and requirements while remaining as generic as possible
[5]. Therefore, the proposed methodology in this work will consist of different
levels of abstraction each having different characteristics and different purposes.
The highest level consists of general phases and stages involved in process mining
projects. This high level view needs to be as generic as possible accounting for
all possible scenarios and contexts process mining can be applied. The lower lev-
els consist of more detailed generic and specific activities for each phase driven
by the context of specific project goals and requirements. The lowest level of
methodology involves an actual run of these activities for a specific project. This
hierarchical nature of the methodology allowing flexibility and addressing the
challenge of balancing genericity and specificity is one of the main features of
the methodology.
In addition, the methodology contains a user guide with best practises, com-
mon approaches and techniques in order to guide the user with various chal-
lenges. The methodology describes the overall approach to extract knowledge
and insight into processes and provide a roadmap to follow while planning and
carrying out process mining projects, addressing two of the process mining chal-
lenges stated in the process mining manifesto [2] namely Improving Usability
for Non-Experts, and Improving Understandability for Non-Experts. It also fa-
cilitates and encourages efforts for automation and reusability of process mining
project flows and currently manual (or partially automated) and time consuming
steps such as data extraction and preparation.
Process mining projects usually involve the following high level phases:
Planning which focuses on both the business and technical aspects of the
project. In this phase project plan, requirements, objectives and available re-
sources are identified and discussed and a concrete project plan is prepared.
Data Discovery containing data extraction and event log preparation. The
journey here, which could be one of the most challenging parts of the project
starts with identifying relevant data sources, finding, extracting, merging and
cleaning the extracted data and leads to preparing an event log in required
formats.
Process Discovery consisting of explorative analysis, process overview and
control-flow discovery. Depending on requirements and the nature of the project,
explorative or goal-driven, the steps taken in this phase vary. Initial insights and
statistics into the process is gained which assist the following step of analysis.
based on this initial insights project might go back to previous phases to modify
and adjust plans or to collect additional data or to adopt different views on the
data.
12 Kiarash Diba
Analysis focusing on the main analysis and evaluation of the result. Dif-
ferent types of analysis can be performed and process mining techniques can
be combined with data mining, statistics and other types of analysis to provide
useful insight and knowledge and address the project objectives.
Knowledge Transfer phase which can be reporting diagnostics and im-
provement insights and/or preparing a monitoring system for operational sup-
port. Due to the iterative nature of process mining projects, there should be
multiple iterations introduced between different phases.
5 Conclusion
This paper outlines an overview and the landscape of a comprehensive method-
ology for process mining. The prospective methodology will involve several hi-
erarchical levels, consisting of high level phases to specific activities for each
phase. In addition, a user guide and best practises and techniques will be in-
cluded to facilitate successful projects in various settings. Continuous extension,
refinement and evaluation need to be performed before and after establishment
of the methodology to ensure generality, completeness and applicability.
References
1. van der Aalst, W.M.P.: Process Mining: Discovery, Conformance and Enhancement
of Business Processes. Springer (2011)
2. van Der Aalst, W.M.P, Adriansyah, A., De Medeiros, A.K.A., Arcieri, F., Baier,
T., Blickle, T., Bose, J.C., Van Den Brand, P., Brandtjen, R., Buijs, J. and Burat-
tin, A.: Process mining manifesto. In International Conference on Business Process
Management, pp. 169-194. Springer, Berlin, Heidelberg (2011).
3. Bozkaya, M., Gabriels, J., Werf, J.: Process diagnostics: a method based on pro-
cess mining. In: International Conference on Information, Process, and Knowledge
Management, eKNOW 2009, pp. 2227. IEEE (2009)
4. van Eck, M.L., Lu, X., Leemans, S.J. and van der Aalst, W.M.: PM 2 : A Process
Mining Project Methodology. In International Conference on Advanced Information
Systems Engineering, pp. 297-313. Springer, Cham (2015).
5. Irny, S.I. and Rose, A.A.: Designing a Strategic Information Systems Planning
Methodology for Malaysian Institutes of Higher Learning (isp- ipta), Information
System 5(1), (2005).
6. Rebuge, A., Ferreira, D.R.: Business Process Analysis in Healthcare Environments:
a Methodology based on Process Mining. Information Systems 37(2), 99116 (2012)
7. Shearer, C.: The crisp-dm model: the new blueprint for data mining. Journal of data
warehousing 5(4), 1322 (2000)