=Paper=
{{Paper
|id=Vol-2767/paper07
|storemode=property
|title=Towards the Identification of Process Anti-Patterns in Enterprise Architecture Models
|pdfUrl=https://ceur-ws.org/Vol-2767/06-QuASoQ-2020.pdf
|volume=Vol-2767
|authors=Barry-Detlef Lehmann,Peter Alexander,Horst Lichter,Simon Hacks
|dblpUrl=https://dblp.org/rec/conf/apsec/LehmannALH20
}}
==Towards the Identification of Process Anti-Patterns in Enterprise Architecture Models==
https://ceur-ws.org/Vol-2767/06-QuASoQ-2020.pdf
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
Towards the Identification of Process Anti-Patterns in
Enterprise Architecture Models
Barry-Detlef Lehmanna , Peter Alexandera , Horst Lichtera and Simon Hacksb
a
RWTH Aachen University, Research Group Software Construction, Aachen, Germany
b
KTH Royal Institute of Technology, Network and Systems Engineering, Stockholm, Sweden
Abstract
IT processes constitute the backbone of an integrated enterprise architecture (EA). The model thereof sustains the develop-
ment and management of the EA. Nevertheless, the quality of such models tends to degrade over time due to, e.g. improper
modeling practices or ineffective evaluation. In this regard, the knowledge of relevant modeling anti-patterns can help iden-
tify, mitigate, and prevent the occurrence of sub-optimal or adverse constructs in the model. In the field of business process
modeling (BPM), a plethora of BPM anti-patterns has been defined and compiled in various taxonomies. However, these BPM
anti-patterns mostly focus on technical issues, which thus are applicable for evaluating workflows but not EA-level processes.
We strongly argue that the concept of process anti-pattern in EA domain can facilitate EA analyses on process-related issues.
To address this gap, this paper presents a catalogue of 18 EA process modeling anti-patterns, which we derived from the
existing BPM anti-patterns. Our result should serve as food for thought and motivation for future research in this context.
Keywords
enterprise architecture, process anti-pattern, model quality
1. Introduction text of EA modeling, specifically the application thereof
in the evaluation of EA models.
IT processes transform fragmented capabilities within an In the field of business process modeling (BPM) re-
enterprise architecture (EA) into consolidated business search, a plethora of BPM anti-pattern taxonomies have
assets. The models thereof often need to be consulted or been proposed [1, 3, 4]. However, these BPM anti-
even adapted in the efforts of managing and evolving the patterns mostly address rather technical aspects like the
EA. Nevertheless, these process models are often devel- use of syntax or layout in the process model, which are
oped with less consideration of quality due to, e.g. time very specific to the modeling notations in use. Moreover,
pressure, little awareness of good modeling practices, or discussions of BPM anti-patterns have mostly been pre-
inadequate evaluation of the models. The uncontrolled sented in workflow modeling notations [5] (e.g. BPMN).
development in this manner will eventually render the These situations hinder the application of BPM anti-
process models useless or even misleading [1]. This sit- pattern to EA practices, in which processes are viewed
uation may hamper the sustainability of EA practices from rather strategic perspectives and modeled in EA
within the organization. modeling notations. We strongly argue that transferring
To avoid this, the development and evaluation of pro- the existing BPM anti-patterns into process anti-patterns
cess models must be guided by the knowledge of relevant in the domain of EA can help improve the development
patterns and anti-patterns. In general, the modeling pat- of processes and their quality that underlie the EA.
tern is defined as a proven solution to a recurring model- The effort of transferring an existing concept into the
ing problem whereas the modeling anti-pattern is defined domain of EA is not new. Salentin and Hacks introduced
as a modeling solution that is known to pose risks [2]. the concept of EA smell, which is defined as a hint to
The understanding of these concepts can help identify, a bad habit that impairs the quality of the EA [6]. In
mitigate, and prevent the occurrence of sub-optimal or their work, they transferred the concept of code smell
adverse constructs within the models [1]. In this study, into the context of EA through conceptual derivation and
we focus on the concept of process anti-pattern in the con- transformation methodology. Inspired by their work, the
same methodology is applied in this study to answer the
QuASoQ 2020: 8th International Workshop on Quantitative following research question (RQ):
Approaches to Software Quality, December 01, 2020, Singapore
" barry.lehmann@rwth-aachen.de (B. Lehmann); RQ What process anti-patterns can be defined to sup-
alexander@swc.rwth-aachen.de (P. Alexander);
port EA modeling activities through the analysis
lichter@swc.rwth-aachen.de (H. Lichter); shacks@kth.se
(S. Hacks) of published process anti-patterns?
0000-0001-6534-278X (P. Alexander); 0000-0002-3440-1238
(H. Lichter); 0000-0003-0478-9347 (S. Hacks) The remainder of this paper is structured as follows:
© 2020 Copyright for this paper by its authors. Use permitted under Creative
Commons License Attribution 4.0 International (CC BY 4.0).
Section 2 gives an overview of previous studies on anti-
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073 CEUR Workshop Proceedings (CEUR-WS.org) pattern or other related concepts (e.g. smell) in the fields
47
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
of BPM and EA modeling research; section 3 describes our the authors provide a catalog of 45 EA smells that origi-
methodology for obtaining process anti-patterns for EA nated from code smells. In their approach, they transform
modeling problems; section 4 elaborates our findings and a catalog of well-known code smells into EA smells and
the analysis thereof; section 6 demonstrates our results categorize the EA smells based on the three concerns of
and discusses the implications as well as threats to the EA: business, application, and technology. Furthermore,
validity thereof; and section 7 motivates future research they present a tool that can detect 14 EA smells. As an
directions and concludes this paper. extension to their work, this study explores the current
knowledge about process anti-pattern to obtain a new
understanding thereof in the EA domain.
2. Related Work Finally, the idea of looking at different abstraction lev-
els in processes to address different stakeholders is not
The concept of anti-pattern was coined in 1995 by Koenig
new. Several studies have been conducted to decom-
[7] to describe a common solution to a recurring problem
pose processes into different abstraction levels. Giachetti
which poses risks of being counterproductive. Although
proposes to divide the process hierarchy into functions,
an anti-pattern may serve as a practical short-term solu-
processes, sub-processes, activities, and tasks [5]. This
tion, the use of it sets a context in which certain changes
study argues that the natural hierarchical attribute of
may become more expensive or impossible. The (unin-
the process should be used. Viljoen, on the other hand,
tentional) use of anti-pattern is highly influenced by, e.g.
decomposes it into enterprise model, macro, business
time pressure, inadequate knowledge of best practices,
process, sub-process, activity, and task [13]. Koschmider
or unforeseen changes.
and Blanchard propose a semiautomatic detection for dif-
The spectrum of studies about anti-patterns covers a
ferent process abstraction levels with processes modeled
wide-range of software engineering topics, such as soft-
with Petri Nets [14]. Their goal is to detect a process
ware development and modeling. In the domain of BPM,
hierarchy in a process model. All these studies com-
a number of taxonomies of process anti-patterns have
monly advocate that the levels of abstraction applied to
been proposed, each of the taxonomies addresses a spe-
processes should meet the goals of relevant stakeholders.
cific area of concern. In 2019, a bibliography of all these
taxonomies was published based on a literature review
study [1]. Therein, the collected taxonomies are divided 3. Methodology
into seven categories based on the addressed modeling
problems. Furthermore, this study suggests several rules To transfer the existing knowledge about process anti-
of thumb in documenting process anti-patterns. The au- pattern into the EA domain, this study follows the
thors of this study advocate the use of this bibliography methodologies proposed by Peffers et al. and Hevner
in the efforts to increase the quality of BP models. et al. for the Design Science Research (DSR) [15], [16]:
Analogous with the concept of process anti-pattern, a type of research which aims to devise an artifact that
the concept of process anomaly is also known in BPM addresses a "heretofore unsolved and important business
research. Vidacic and Strahonja present a literature problem" by drawing on the existing knowledge. The
review of this concept in which the collected process resulting artifact must be rigorously evaluated in terms
anomalies are divided into three categories: structural, of its "utility, quality, and efficacy" and effectively com-
semantical, and syntactical anomalies [8]. They also sug- municated to the relevant audience.
gest approaches to the mitigation or prevention thereof. This study is performed as follows: At first, we col-
Suchenia et al. provide a brief overview of BPM anti- lected knowledge about the 336 already-published BPM
patterns, present these in BPMN models, and categorize anti-patterns in scientific literature, which have been
these into three categories: syntactic, structural, and compiled in [1] and are publicly accessible on [17]. Based
control flow anti-patterns [9]. Trcka et al. present data- on a mapping of modeling notations between BPMN and
flow anti-patterns and an approach to identifying these ArchiMate, we processed these BPM anti-patterns and
[10]. Further in this topic, Sadiq et al. identify seven finally derived 18 process anti-patterns of EA relevance,
common data-flow anti-patterns and provide the basic which are then documented in a structured template and
algorithm to address these [11]. Finally, Döhring and exemplified in ArchiMate models. A closer look into this
Heublein present a taxonomy of control-flow, rule-based, procedure is provided in the following subsections.
and data-flow anti-patterns; demonstrate examples of
such anti-patterns in BPMN; and suggest detection as 3.1. Notation Mapping
well as prevention mechanisms thereof [12].
In the domain of EA modeling, the concept of anti- Most BPM anti-patterns have been analyzed and visu-
pattern remained unknown until the recent suggestion alized in BPMN. While BPMN provides a full-fledged
of an EA smell taxonomy by Salentin and Hacks. Therein, framework to create graphical business processes mod-
48
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
BPMN ArchiMate Source
Business Process Diagram, Pool, Lanes Process [21, 19]
Activities, Task, Sub-Process Function [21]
Collaboration Diagram Interaction [21]
Event Event [21, 18, 19]
Data Object Object [21, 19]
Lane Business Role, Business Actor, Application Component [21, 19]
Sequence flow Triggering, Flow [20]
Data association Access [20]
Inclusive and parallel gateways And-Junction [20, 18]
Exclusive and event-based gateways Or-Junction [20, 18]
Table 1
Mapping from BPMN elements to ArchiMate
els, BPMN is intended for rather detailed business process the various situations of deadlock are addressed by differ-
modeling, such as the modeling of conversation, chore- ent BPM anti-patterns. While such specialized modeling
ography, and collaboration models. However, such busi- problems are indeed identifiable within BPMN models,
ness process models constitute only a small area within such detailed problems are not visible through the high-
the broad view of EA. Therein, processes are analyzed level notations of ArchiMate. Therefore, we pruned such
from rather strategic perspectives, e.g. the connection detailed branches of classification within the selected
of all high-level processes to the surrounding organiza- taxonomies, thereby reducing the size of our analysis to
tion units to achieve strategic targets. The modeling of 200 BPM anti-patterns.
such perspectives has been facilitated by a number of EA Next, this study follows a defined procedure to derive
modeling languages; the most popular one is ArchiMate. process anti-patterns of EA relevance, which is as follows:
We strongly argue that method supports for analyzing Firstly, BPMN elements found in the names and descrip-
EA processes must be built on top of an EA modeling tions of the BPM anti-patterns are translated into Archi-
language. Therefore, to transfer the existing BPM anti- Mate elements based on the notation mapping shown in
patterns into the EA domain, we first need to create a table 1. Secondly, each translated description is evalu-
mapping between BPMN and ArchiMate to figure out ated to determine whether it is comprehensible within
the possibilities of deriving something of EA relevance. the context of EA. Through this step, we derived 18 pro-
Previous studies have suggested several mappings be- cess anti-patterns of EA relevance, while excluding the
tween BPMN and ArchiMate [18, 19, 20, 21]. They show rest because the translated descriptions thereof do not de-
that both modeling languages share some conceptual scribe relevant EA modeling problems (e.g. anti-patterns
similarities. Firstly, both provide similar notations for related to syntax errors such as the sequence flow crosses
connecting process elements (e.g. the sequence, default, process boundary [3]). Among the derived process anti-
and conditional flows) and regulating these with gate- patterns, some are directly applicable in the EA domain
ways or junctions. Secondly, both support the modeling only after applying the notation mapping (e.g. the Layout
of similar relationships between processes and the related Deficit [1]), while some other require broad modification
elements [19]. For example, BPMN’s support for creating in the description thereof to depict valid EA modeling
relationships between e.g. activities and data objects is problems (e.g. Useless Test).
similar to ArchiMate’s support for creating relationships
between e.g. business processes and business objects. In
table 1, we list the mappings of notations which we use 4. An initial catalogue of process
as a basis to conduct the next steps. anti-patterns in EA
3.2. Transformation Design To support the usability of our contribution, we have
organized the resulting 18 process anti-patterns in a cata-
From our analysis of related studies in BPM research, logue, which is publicly accessible on our [22]. Since this
we found several taxonomies which collect in total of catalogue is still in its initial stages, we encourage further
336 BPM anti-patterns and classify these based on the extensions and improvements to it. A detailed discussion
characteristics thereof [1] [17]. From the perspective of on future research directions is given in section 7.
our study, we argue that the depth of these taxonomies In this section, we first introduce the categorization
stretches beyond the context of EA analysis. For example, used in the catalogue. Next, we provide a deeper look
49
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
into one process anti-pattern under each category. Lastly, Table 2
we describe the template used to document some general Categories of EA anti-pattern
attributes of the process anti-patterns. Category Anti-pattern
Semantic Error End event missing
4.1. Categorizing process anti-patterns Start event missing
Control-flow Dead Element
To present the proposed catalogue in an organized and Problem Deadlock
systematic manner, we divide the contained process anti- Infinite loop
patterns into the following five categories that we derived Lack of synchronization
from some selected categories in [1]. Undefined junction condition
The category of semantic error includes process anti- Understand- Junction named as element
patterns that address the semantically error or inconsis- ability Problem Layout deficit
tent parts of the model under analysis. These problems Language deficit
Missing negative case
are commonly caused by the improper or deficient use of
The word And in element name
modeling notations, which may distort the understanding The word Or in element name
of the model. It is worth to mention that such semantic Useless test
errors should not be confused with syntax errors as the Rule-related Contradiction in input
latter addresses the violation against the rules of assem- Defect
bling the notations, whereas the earlier addresses the Data-flow Inconsistent data
false or ambiguous impression conveyed by the model related Defect Mismatched data
[3]. Due to the gap between the syntactic rules of BPMN Missing data
and ArchiMate, we derived no process anti-patterns of
EA relevance under the syntax error category.
The category of control-flow problem includes pro- Furthermore, to give a closer look into the catalogue, we
cess anti-patterns that address the flawed concurrency of elaborate one process anti-pattern under each category
process flows which raises unpredictability in the final and provide an example thereof.
outcome. Such issue occurs when the modeled processes Under the category of semantic error, the end event
are split or joined without properly considering the in- missing occurs when the modeled process does not
fluence thereof to the final outcome. clearly specify the end events [3] [23], which then causes
The category of understandability problem in- confusion or misinterpretation about the valid conditions
cludes process anti-patterns that address the excessive or to finalize or abort the process execution. An example
lack of complexity within the model, thereby requiring of this problem is when multiple high-level processes of
excessive efforts to understand the process under analy- different organization bodies are integrated within the
sis in its full context. Such an issue is commonly caused EA model without specifying the points when the collab-
by the improper/inconsistent level of information gran- orative outcomes have been achieved or any disruptions
ularity or inadequate/imprecise coverage of the actual have to be handled. A solution to this anti-pattern would
process in reality. be to simply introduce end event elements that clearly
The category of rule-related defect includes process signal all possible ways to end the process.
anti-patterns that address the contradictions among the Under the category of control-flow problem, the lack
rules specified within the model. Such an issue may of synchronization refers to the situation when the
occur when e.g. the actual process is not well-defined, modeled process does not specify a proper synchroniza-
well-communicated, or well-understood among some tion among concurrent process flows, thereby showing
contributors to the rules specification in the model. no predictable outcome. An example of such a situation
The category of data-flow related defect includes is when the modeled process is split by an AND junction
process anti-patterns that address the proneness to con- and later joined by an OR junction [24, 25]. A solution to
flicts when the same data object is concurrently used this anti-pattern is to ensure the highest test coverage of
for different kinds of transactions. These issues are com- all points of synchronization specified in the model.
monly caused by the overlapping data responsibilities Under the category of understandability problem, the
among different processes or the centralization of too useless test is identified when the modeled process ful-
many data in a single data object. fills only some of all the possible cases in reality [26],
thereby making it impossible to identify and test the
4.2. Demonstrating process anti-patterns real extent of the supported problem domain. An exam-
ple thereof is when the handling of possible mistakes or
Next, we name the 18 process anti-patterns and map these disruption during the process execution is not specified
under the aforementioned categories, as listed in table 2. within the model. As a solution, the model should be
50
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
Attribute Meaning
Name Gives the anti-pattern a meaningful designator
Problem Describes why the anti-pattern leads to problems
Consequences Describes what the consequences of the anti-pattern are
Solution Describes a solution to the anti-pattern
Graphical Definition Shows a graphical representation of the anti-pattern to re-
duce misinterpretations
Table 3
Documentation attributes of EA process anti-pattern (adopted from [1]).
incrementally and iteratively developed along with the 5. Applying process anti-patterns
continuous identification of relevant test cases until it
reaches a reasonable level of complexity and covers the
in EA
complete problem domain [27]. In order to illustrate the concept and the usage of process
Under the category of rule-related defect, the contra- anti-pattern in the context of EA, we analyze a slightly
diction in input occurs when some rules applied to the modified ArchiMate EA model and annotate it with anti-
modeled process may contradict with each other, thereby pattern information. The model used, depicted in fig. 1,
hindering the process execution to continue as intended is contained in a publicly accessible collection of Archi-
[12]. An example thereof is when a certain data object Mate example EA models [30]. This model defines how
passes the input validations specified on a junction de- new orders are processed. After a new order is received,
spite being actually invalid for the supplied process. A planning the order and evaluating the customers’ credi-
solution to this anti-pattern is to continuously perform a bility are done in parallel. After an approved proposal is
rigorous combinatorial testing of all possible input types available, the customer signs the respective contract to
and all the rules applied to the modeled process. accept the proposal.
Under the category of data-flow-related defect, the in- When analysing the model, it can be observed that the
consistent data describes the situation in which data processes Evaluate customer credit and Plan order
objects (e.g. customer records or insurance claim) are do not wait for each completion before continuing to the
accessed by concurrent process flows, thereby making it Develop approved proposal process, which may lead
prone to data handling mistakes. An example thereof is to undesired results (e.g. the contract does not consider
when multiple processes work on duplicates of the same the customer’s credibility). These are consequences of
data object, and a (manual) synchronization procedure the lack of synchronization anti-pattern.
between the duplicates is required after every modifica- When analyzing the usage of the Order Data element,
tion on one side. To mitigate this, the strategy of handling
we identify the inconsistent data anti-pattern because
the data must be carefully defined and implemented. this data can be changed without rerunning dependent
processes (e.g. Evaluate customer credit).
4.3. Documenting process anti-patterns Furthermore, the Evaluate customer credit process
exhibits the useless test anti-pattern, as only the positive
In general, the documentation of modeling anti-patterns test result is modeled.
includes many attributes of modeling patterns [28] to- Next, we can identify the anti-pattern end event miss-
gether with some other attributes like cause and detection ing because no clear termination is defined for this EA
[29]. To document the process anti-patterns identified process model. The process could end in Refuse pro-
in this study, we derive some attributes from the tem- posal or in Accept proposal.
plates for documenting BPM anti-patterns introduced in Finally, we detect the contradiction in input anti-
[1], as shown in table 3. Please note that, at the time pattern at the Or-Junction that splits the control flow
of this writing, not all attributes have been completed after Develop approved proposal. There, the incorrect
for each process anti-pattern due to the need of further condition will never lead to the Refuse proposal process
information and analysis. and therefore makes it a dead process.
6. Discussion
Some of the main goals of applying the EA discipline
within an enterprise is to ensure the business-IT align-
51
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
Useless test
Evaluate
customer credit Contradiction in input Refuse
proposal
Process does
New order Develop
Enter order contain end event
received approved proposal
Lack of
synchronization Accept
proposal
Plan order
Contract
Order Data
Customer
signature
Inconsistent data
CRM ERP
Figure 1: An example EA ArchiMate model with annotated process anti-patterns
ment [31] and to develop solid IT strategies that can help patterns in an EA model, as has been drafted in a program
achieve strategic targets [32]. For this reason, this section that currently detects 14 EA smells [6].
seeks to answer the ultimate question of this study: "how
can enterprise architects benefit from the contribution 6.2. Threats to validity
of this study?" In this section, we describe the use of the
proposed process anti-patterns to support both research The results of this study have to be seen in the light of
and practice of EA methods. Following to this, we discuss some limitations. The limitations that affect the results
the threats to the validity of our results. of this paper are the lack of previous research and bias
during the anti-pattern transformation.
Lack of previous research. There is little research
6.1. Implication for researchers &
on transforming low-level process methods to be applica-
practitioners ble for high-level processes and even much less on bring-
The concept of EA debt has recently been introduced as ing together the modeling notations for such processes.
the deviation between the current state and the hypo- In addition, the already suggested mappings between the
thetical ideal state of the enterprise [33]. Factors to such modeling notations for low-level and high-level processes
deviation in EA (e.g. sub-optimal or adverse solution are rather described as informal and lack of theoretical
design) are likely to be identifiable within the EA models foundations, thereby leaving room for interpretation or
created or used during the planning, development, eval- different mapping solutions. This might reduce the va-
uation, or communication of the EA [6]. Therefore, the lidity of our results because our approach relies on the
ability to recognize the existence of such deviation in EA existing mapping between BPMN and ArchiMate.
models is needed, and the concept of process anti-pattern Bias during the anti-pattern transformation. To
proposed in this study is intended to support such ability reduce bias when selecting the relevant BPM anti-
with sharp focus on processes. In this case, practitioners patterns to be transformed, we first need to establish ob-
can use the process anti-patterns catalogue to scan the jective selection criteria. Thus, we defined a mapping of
EA models for potential EA Debt. In any case, it is bene- notations and apply it on the collected BPM anti-patterns
ficial that the practitioners are aware of the potentially to prune the ones that do not fit in the new domain.
vulnerable parts of the EA Model as further development Despite this, subjective assessment is still inevitably in-
could be hindered if these remain ignored [6]. volved during the process, thereby leaving room for in-
Furthermore, we also intend to impart food for thought terpretation and may not produce unique results.
into the EA research community and provide a basis for
further research works in this topic. For example, the
process anti-patterns identified in this study might help
researchers to extend the automatic detection of EA anti-
52
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
7. Conclusion & Future Works tems for a Sharing Society, ECIS 2019, Stockholm
and Uppsala, Sweden, June 8-14, 2019, 2019. URL:
The concept of anti-pattern has been long known to help https://aisel.aisnet.org/ecis2019_rp/157.
recognize common solutions that are not sustainable for [2] M. Fellmann, A. Koschmider, R. Laue,
the future development. However, little emphasis has A. Schoknecht, A. Vetter, Business process model
been put on studying this concept in the context of EA. patterns: State-of-the-art, research classification
The first step in this direction has recently been made to and taxonomy, Business Process Management Jour-
transfer the existing code smells into the EA domain, out nal (2018). doi:10.1108/BPMJ-01-2018-0021.
of which an initial catalogue of 45 EA smells has been de- [3] T. Rozman, G. Polančič, R. V. Horvat, Anal-
veloped and proposed [6]. To pursue a meaningful exten- ysis of most common process modelling mis-
sion to this result, this study focuses on transferring the takes in bpmn process models, in: EuroSPI
existing BPM anti-patterns [1] into process anti-patterns 2007, Industrial Proceedings, 2007, pp. 1.21 – 1.31.
for EA modeling problems, with reference to a mapping URL: https://2020.eurospi.net/images/proceedings/
between the notations of BPMN and ArchiMate. EuroSPI2007-ISBN-978-3-9809145-6-7.pdf.
The process anti-patterns identified in this study are [4] S. Von Stackelberg, S. Putze, J. Mülle, K. Böhm, De-
compiled in a catalogue that is publicly accessible on our tecting data-flow errors in bpmn 2.0, Open Journal
[22]. Therein, the process anti-patterns are categorized of Information Systems (OJIS) 1 (2014) 1–19.
and documented in a well-known template to ease the use [5] R. E. Giachetti, Design of Enterprise Systems: The-
or extension thereof by EA practitioners and researchers. ory, Architecture, and Methods, 1st ed., CRC Press,
The practical use of this catalogue covers a broad range Inc., USA, 2010.
of topics, starting from the identification of flaws in EA [6] J. Salentin, S. Hacks, Towards a catalog of enter-
models to the identification of EA debt. Nevertheless, prise architecture smells, in: N. Gronau, M. Heine,
this catalogue is still in its initial stages. Much more in- H. Krasnova, K. Poustcchi (Eds.), Entwicklun-
formation and analyses are needed before this catalogue gen, Chancen und Herausforderungen der Digi-
can be evaluated in real industrial contexts. Therefore, talisierung: Proceedings der 15. Internationalen
to motivate further research in this topic, the rest of this Tagung Wirtschaftsinformatik, WI 2020, Potsdam,
section outlines some ideas of future research directions. Germany, March 9-11, 2020. Community Tracks,
The future research directions in this topic can be di- GITO Verlag, 2020, pp. 276–290. doi:10.30844/
vided into three main topics: to pursue different methods wi∖_2020∖_y1-salentin.
for defining new EA anti-patterns, to perform empiri- [7] A. Koenig, Patterns and antipatterns, J. Object
cal studies for improving both conceptual and practical Oriented Program. 8 (1995) 46–48.
knowledge in this context, and to develop tool supports [8] T. Vidacic, V. Strahonja, Taxonomy of Anomalies
for the automatic detection of the anti-patterns in EA in Business Process Models, 2014, pp. 283–294.
models. In terms of analyzing more methods to find EA doi:10.1007/978-3-319-07215-9_23.
anti-patterns, we suggest to investigate new domains (e.g. [9] A. Suchenia, T. Potempa, A. Ligęza, K. Jobczyk,
documentation, data, or requirement anti-patterns) to ex- K. Kluza, Selected Approaches Towards Taxon-
tend the catalogue with adapted or new EA anti-patterns. omy of Business Process Anomalies, volume
Also, the investigation of cause-and-effect relationships 658, 2017, pp. 65–85. doi:10.1007/978-3-319-
among the identified EA anti-patterns may provide in- 47208-9_5.
sights on the possible propagation of impacts thereof. In [10] N. Trcka, W. M. P. van der Aalst, N. Sidorova,
terms of empirical studies, evaluations using real EA mod- Data-flow anti-patterns: Discovering data-flow er-
els from different business domains can help to verify and rors in workflows, in: P. van Eck, J. Gordijn,
improve the quality of the catalogue and the proposed R. J. Wieringa (Eds.), Advanced Information Sys-
method supports. Finally, tool supports can be developed tems Engineering, 21st International Conference,
to support the continuous detection of EA anti-patterns CAiSE 2009, Amsterdam, The Netherlands, June 8-
or the (early) signs of their occurrences. 12, 2009. Proceedings, volume 5565 of Lecture Notes
in Computer Science, Springer, 2009, pp. 425–439.
doi:10.1007/978-3-642-02144-2∖_34.
References [11] S. W. Sadiq, M. E. Orlowska, W. Sadiq, C. Foulger,
[1] A. Koschmider, R. Laue, M. Fellmann, Business Data flow and validation in workflow modelling,
process model anti-patterns: a bibliography and in: K. Schewe, H. E. Williams (Eds.), Database Tech-
taxonomy of published work, in: J. vom Brocke, nologies 2004, Proceedings of the Fifteenth Aus-
S. Gregor, O. Müller (Eds.), 27th European Confer- tralasian Database Conference, ADC 2004, Dunedin,
ence on Information Systems - Information Sys- New Zealand, 18-22 January 2004, volume 27 of
CRPIT, Australian Computer Society, 2004, pp. 207–
53
8th International Workshop on Quantitative Approaches to Software Quality (QuASoQ 2020)
214. URL: http://crpit.scem.westernsydney.edu.au/ [23] C. Dias, V. Stein Dani, J. Mendling, L. Thom,
abstracts/CRPITV27Sadiq.html. Anti-patterns for Process Modeling Problems: An
[12] M. Döhring, S. Heublein, Anomalies in rule-adapted Analysis of BPMN 2.0-Based Tools Behavior, 2019,
workflows - a taxonomy and solutions for vbpmn, pp. 745–757. doi:10.1007/978-3-030-37453-
in: 2012 16th European Conference on Software 2_59.
Maintenance and Reengineering, 2012, pp. 117–126. [24] Z. Han, P. Gong, L. Zhang, J. Ling, W. Huang, Defi-
[13] S. Viljoen, Reflections on business process nition and detection of control-flow anti-patterns
levelling, white paper, 2012. URL: https: in process models, in: 2013 IEEE 37th Interna-
//realirm.com/sites/default/files/whitepapers/ tional Computer Software and Applications Con-
reflections_on_business_process_leveling_0.pdf. ference Workshops (COMPSACW), IEEE Computer
[14] A. Koschmider, E. Blanchard, User assistance for Society, Los Alamitos, CA, USA, 2013, pp. 433–438.
business process model decomposition, in: Proceed- doi:10.1109/COMPSACW.2013.111.
ings of the 1st IEEE International Conference on [25] R. Laue, A. Awad, Visualization of business pro-
Research Challenges in Information Science, 2007, cess modeling anti patterns, ECEASST 25 (2010).
pp. 445–454. doi:10.14279/tuj.eceasst.25.344.
[15] K. Peffers, T. Tuunanen, M. Rothenberger, S. Chat- [26] R. Laue, W. Koop, V. Gruhn, Indicators for open
terjee, A design science research methodology issues in business process models, in: International
for information systems research 24 (2007) 45–77. Working Conference on Requirements Engineering:
doi:10.2753/MIS0742-1222240302. Foundation for Software Quality, Springer, 2016, pp.
[16] A. R. Hevner, S. T. March, J. Park, S. Ram, Design 102–116.
science in information systems research, MIS Q. 28 [27] P. Desfray, G. Raymond, Chapter 5 - key mod-
(2004) 75–105. URL: http://misq.org/design-science- eling techniques, in: P. Desfray, G. Raymond
in-information-systems-research.html. (Eds.), Modeling Enterprise Architecture with TO-
[17] Business process model patterns classification, GAF, The MK/OMG Press, Morgan Kaufmann,
15.11.2020. URL: http://www.bpmpatterns.org/. Boston, 2014, pp. 67 – 91. doi:https://doi.org/
[18] Archimate® 3.1 specification, 27.06.2020. 10.1016/B978-0-12-419984-2.00005-7.
URL: https://pubs.opengroup.org/architecture/ [28] E. Gamma, R. Helm, R. Johnson, J. Vlissides, Design
archimate3-doc/. Patterns, volume 47 of Addison Wesley Professional
[19] D. Orlovskyi, A. Kopp, Enterprise architecture Computing Series, 1995.
modeling support based on data extraction from [29] J. Bogner, T. Boceck, M. Popp, D. Tschechlov,
business process models, in: S. Subbotin (Ed.), S. Wagner, A. Zimmermann, Towards a collabo-
Proceedings of The Third International Workshop rative repository for the documentation of service-
on Computer Modeling and Intelligent Systems based antipatterns and bad smells, in: 2019
(CMIS-2020), Zaporizhzhia, Ukraine, April 27-May IEEE International Conference on Software Ar-
1, 2020, volume 2608 of CEUR Workshop Proceed- chitecture Companion (ICSA-C), 2019, pp. 95–101.
ings, CEUR-WS.org, 2020, pp. 499–513. URL: http: doi:10.1109/ICSA-C.2019.00025.
//ceur-ws.org/Vol-2608/paper38.pdf. [30] P. Beauvoir, A collection of archimate® models.,
[20] M. Lankhorst, Combining archimate® 3.0 27.06.2020. URL: https://github.com/archimatetool/
with other standards – bpmn, 6.9.2016. ArchiModels.
URL: https://bizzdesign.com/blog/combining- [31] J. Gøtze, The changing role of the enterprise archi-
archimate-3-0-with-other-standards-bpmn/. tect, in: 2013 17th IEEE International Enterprise Dis-
[21] L. Penicina, Linking bpmn, archimate, and BWW: tributed Object Computing Conference Workshops,
perfect match for complete and lawful business 2013, pp. 319–326. doi:10.1109/EDOCW.2013.42.
process models?, in: J. Grabis, M. Kirikova, [32] C. Strano, Q. Rehmani, The role of the enterprise
J. Zdravkovic, J. Stirna (Eds.), Short Paper Pro- architect, Inf. Syst. E-Business Management 5 (2007)
ceedings of the 6th IFIP WG 8.1 Working Confer- 379–396. doi:10.1007/s10257-007-0053-1.
ence on the Practice of Enterprise Modeling (PoEM [33] S. Hacks, H. Höfert, J. Salentin, Y. C. Yeong,
2013), Riga, Latvia, November 6-7, 2013, volume H. Lichter, Towards the definition of enterprise
1023 of CEUR Workshop Proceedings, CEUR-WS.org, architecture debts, in: 23rd IEEE International
2013, pp. 156–165. URL: http://ceur-ws.org/Vol- Enterprise Distributed Object Computing Work-
1023/paper15.pdf. shop, EDOC Workshops 2019, Paris, France, Oc-
[22] B.-D. Lehmann, Enterprise process anti- tober 28-31, 2019, 2019, pp. 9–16. doi:10.1109/
pattern, 05.10.2020. URL: https://ea-anti- EDOCW.2019.00016.
pattern.pages.rwth-aachen.de/process-anti-
patterns/.
54