=Paper=
{{Paper
|id=Vol-2793/paper7
|storemode=property
|title=Panel Discussion: Enterprise Modeling in the Digital Age
|pdfUrl=https://ceur-ws.org/Vol-2793/paper7.pdf
|volume=Vol-2793
|authors=Dirk van der Linden,Jolita Ralyté,Kurt Sandkuhl,Jelena Zdravkovic
|dblpUrl=https://dblp.org/rec/conf/ifip8-1/LindenRSZ20
}}
==Panel Discussion: Enterprise Modeling in the Digital Age==
https://ceur-ws.org/Vol-2793/paper7.pdf
Panel Discussion: Enterprise Modeling in the Digital Age
Dirk van der Lindena, Jolita Ralytéb, Kurt Sandkuhlc, and Jelena Zdravkovicd
a
Northumbria University, Department of Computer and Information Sciences, Newcastle upon Tyne, UK
b
University of Geneva, Geneva, Institute of Information Service Science, Geneva, Switzerland
c
University of Rostock, Business information systems, Rostock, Germany
d
Stockholm University, Department of Computer and Systems Sciences, Kista, Sweden
Abstract
This paper reports the discussion of the “Enterprise Modelling in Digital Age” plenary panel
held at the PoEM2020 conference. The panel was concerned with the interplay between
emergent digitalization, big data, high complexity, needs for competencies and collaboration,
in the scope of enterprise modelling methods and enterprise architecture frameworks. Both in
terms how these global organisations related phenomena provide new challenges to be included
in enterprise modelling, as well as of how they can benefit from the modeling.
Keywords 1
Enterprise Modeling, Enterprise Architecture, Digitalisation, Big Data, Capability
1. Introduction
Enterprise Modeling (EM) has been established as an approach for the development of IT solutions
congruent with the business of their organizations [1]. By a compound modeling of business goals,
services, processes, resources, actors, IS capabilities and requirements, EM describes a desired state of
the organization integrating business and technology perspectives [2]. The notion closely relates to
Enterprise Architecture (EA), which applies various architecture principles through model-based
frameworks for guiding organizations towards business and technology changes necessary to execute
their strategies.
Ongoing digital transformation, technological disruptions, and business agility require enterprise
modeling to evolve to keep up with the pace of change, but moreover - to predict, facilitate, and leverage
that change to the advantage of the business. Today’s actual concerns in IS engineering relate to high
needs for business integration, setting demands to efficient deployment, connectivity, interoperability,
and management of supporting digital components that are of diverse structure, size and dynamics. EM
and EA frameworks are increasingly moving the modeling focus from traditional company oriented
towards ecosystems orchestrated on digital platforms [3]. Today’s business environments entail
cooperation. Because the pace of activities across business sectors accelerates, constellations are
becoming more complex. This means that, to survive and grow, business organizations need to reshape
the methods and models for their IS design to consider both internal and external relationships, as well
as to by digital means accurately plan, analyze, simulate and integrate desired capabilities; and to
facilitate their resilient evolution [4].
The aim of this panel discussion was to explore the challenges and opportunities that massive
digitalization of business data and functions brings to the field of EM&EA:
How modeling frameworks need to change and extend their concepts to meet the needs for
modeling networked organizations in terms of digital business ecosystems?
How existing modeling methods need to be extended to involve and automate acquisition of
massive and diverse digital data and functionalities?
PoEM’20 Forum: 13th IFIP WG 8.1 Working Conference on the Practice of Enterprise Modelling, Forum, November 25–27, 2020, Riga,
Latvia
EMAIL: djt.vanderlinden@gmail.com (D. van der Linden); jolita.ralyte@unige.ch (J. Ralyté); kurt.sandkuhl@uni-rostock.de (K. Sandkuhl);
jelenaz@dsv.su.se (J. Zdravkovic)
ORCID: 0000-0002-8597-3156 (D. van der Linden); 0000-0001-8561-3567 (J. Ralyté); 0000-0002-7431-8412 (K. Sandkuhl);
0000-0002-0870-0330 (J. Zdravkovic)
© 2020 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
How can we increase the efficiency of enterprise modeling, given the current and future
business dynamics and interorganizational scopes, to facilitate timely deployment of IS
solutions?
The panel involved Dirk van der Linden, Jolita Ralyté, and Kurt Sandkuhl as the panelists, while
Jelena Zdravkovic was the session chair. The remainder of the paper provides the views of the panelists
regarding the three main questions described above: section 1 focuses on the ability of modeling
frameworks for supporting objectivity, with a focus on social and cultural perspectives; section 2
discusses possible means for enterprise modeling methods to systematically cope with context-
awareness and big data; and section 3 takes the focus to discussing the challenge of efficiency of
enterprise modeling in respect to business complexity, dynamics, and needed humans’ modeling
competencies and collaboration. The conclusion reflects the positions presented in the previous session,
as well as it summarizes the outcome of the panel discussion.
2. Modeling frameworks need to stop treating all model entities as neutral
There is a plethora of modeling languages to capture everything possibly relevant to an enterprise. The
major modeling languages and frameworks, like, e.g., ArchiMate, similarly, have access to numerous
extensions to further build out what they can capture and integrate. Yet, a key thing still seems to elude
our grasp, even though it has become increasingly critical in this age of data-driven interactions and
ever-faster spread of information. It is something seemingly trivial, that might even be done by simple
annotations: ensuring that key concepts which business processes rely upon are not treated as if they
are some neutral entities avoid of inherent bias. In particular, “resources” in models (whether physical
things like servers, or non-physical things like datasets and algorithms) should be explicitly handled as
inherently biased entities. This is vital because systems (whether it be some piece of hardware, software,
a business model, or indeed an entire enterprise), and the way we design them, reinforce the biases and
viewpoints of its designers [5], and if left unchecked, can reinforce systemic oppression and deepen
social inequity [6]. Yet, resources often seem to be seen as neutral black-boxes, devoid of principle or
value, and positively contributing to capabilities and agency, rather than potentially also taking away
from it.
Consider, a simple example. An enterprise is looking to optimize access procedures to its premises
and, wanting to do away with human porters, opts to automate matters. They chose to use facial
recognition technology. This can be integrated into an enterprise model fairly trivially by considering
the different kinds of resources needed for the similar “check employee is authorized” functionality that
already exists. That would likely involve the physical technologies (e.g., cameras, network
infrastructure, and so on), and the needed application resources (e.g., facial recognition software). But
the datasets this software are trained on have been shown to have significant bias [7] and lead to high
accuracy for white men. Anything else, and accuracy drops. Practically, while white male employees
might easily pass this new access procedure, everyone else would suffer. This could have been trivially
caught if only we systematically ensured that resources in our models have to make explicit their
assumptions and biases—here that the resource was trained only on a specific ethnic group.
Consider another example. An AgriTech company develops a piece of sensor-driven hardware that
works perfectly in the context of their professional culture, and, through a careful balance of human
behaviors and expectancies of the sector in that country, come together to mitigate, indeed nearly
nullify, the impact of the most prevalent cyber threats [8]. That mitigation is not borne out of the
technology itself, but relies on the complex, situated nature in which it exists. Yet, in reality, such
technology rapidly diffuses across borders, and is adopted and integrated in new enterprises, blissfully
unaware of how dangerous it might be to them given their situated particularities. Again, this could
have been trivially caught if only we systematically ensured that resources in our models have to make
explicit their assumptions and biases—here that the technology is assumed to operate safely only in a
specific cultural situated context.
One might argue that little extension of modeling languages or frameworks is needed per se. Rather,
we can rely on our Enterprise Architecture principles, or the use of explicit goal models and other
requirements specifications that tell us not to make biased decisions, or even simply not to use biased
resources. But therein lies the crux—Enterprise Modelers will need to know when a model, or even
65
just an entity within it, violate those principles, or goals, or requirements. And to get to that point,
whether manually or automated, we ought to make sure the way we model forces ourselves to ensure
we do not treat key concepts as being neutral. Even if that only means annotating their biases and
assumptions for now. Given the growing trend of using large-scale datasets and automated reasoning
to inform enterprise decision-making, it is ever more imperative to give, nay, tell, modelers to model
explicitly the bias inherent in our models and stop treating them as neutral.
3. What is needed to manage diverse digital data and functionalities?
One of the most obvious features of the digital age probably is the massive and increasing use of digital
data. Increased availability of computing power at low costs, growing digitalization of most industrial
and service domains and new data-intensive business models are further increasing this trend. Although
enterprise modeling methods and languages in general are well prepared or even explicitly designed for
analyzing, capturing or designing data and information processing solutions, there are some aspects of
the growing data intensity that deserve discussion with respect to potential implications on EM. Smart
connected products and context-aware business services are among the areas that require high volumes
of data and can serve as examples to discuss implications on EM.
Smart connected products basically are physical products enhanced by built-in sensors and actuators,
embedded software modules and connectivity components connecting them via the Internet to their
owners or producers [9]. This allows for monitoring and control of the product but also for new kinds
of services, such as performance tuning, energy optimization or integration of third-party applications.
Examples are smart cargo boxes, connected cars or intelligent household devices. Every device creates
a substantial amount of data by reporting relevant events to its owner, operator or producer (e.g.
activities, error situation, configuration changes, etc.). As long as all possible event types are known or
can be designed during a modeling project, EM can be done with existing methods and languages. But
new kinds of event types or changing ones caused by integration of smart connected products into
business models that emerge during use requires dealing with incompleteness of the modeling domains
or uncertainty what events to handle and how to handle them. Established EM languages and methods
are in general not prepared for this challenge.
Context-based services are designed to adapt to changes in their application environment, which
requires continuous evaluation of indicators that in turn are usually based on data streams from sensors,
filtering of transaction data or analysis of profiles and behavior. Examples are compliance monitoring
applications in different regulative contexts or assistive systems for individual users requiring situation
recognition and adaptation. Here, the challenge for EM lies in high complexity mutual dependency of
indicators and high variability of the overall system. Using EM techniques to model such context-based
systems is possible (as shown, for example, in capability-as-a-service project [10]), but adaptation often
is expensive in terms of modeling effort. Modeling the adaptation mechanisms might reach the same
limits of incompleteness and dealing with uncertainty as mentioned for the smart connected products.
The impact of data-intensive applications areas on EM depends on the purpose of modeling. In EM
projects aiming for analysis of the current situation in an enterprise, preparation of decision making for
new business models or the design of organizational change, modelling the specifics of event types or
context settings is not required with all technical details. The focus is more on business processes,
organizational structures or product/service structures, which is possible with established EM methods
and languages. However, when EM is applied as an activity in a software system engineering process,
specification of input and output data, information models and information flow as complete and exact
as possible would be required to allow for model-based engineering scenarios. In this case, we should
investigate “plug-in” concepts for EM meta-models, dynamic extensions of modeling languages or
generation of models from event or data streams. Here, not only organizational change has to be
addressed but also the possibility to define policies, frame conditions or business rules as manifestations
of the desired business logic.
Data-intensive application areas also raise the question if enterprise models should continue to be
primarily design time constructs or rather be developed towards runtime artefacts. “Design-time”
construct means that most enterprise models only capture the design of (organizational) solutions, but
are not executable or enactable. Such models either serve as blueprint or documentation of the “to-be”
66
situation and are used by human actors or they have to be transformed into executable formats including
operational semantics. “Runtime-artefact” would mean that an enterprise model could even be the basis
for execution, which not only requires more formal specification but also would allow to actually handle
the required extensions mentioned above.
4. Towards efficient enterprise modeling
Enterprise modeling takes time and effort. But, once produced, models have a very limited lifespan as
they are mare blueprints of a time-stamped situation, current (as-is) or desired (to-e) one. This situation
does not last. Digital transformation takes us to a society in which things are increasingly dynamic.
Organizations are constantly undergoing change that has an impact at all levels of EA: business
structure and processes, services and products, information systems and technology. A model that is
valid today may already be obsolete tomorrow. Does it then still make sense to build the models? Can
we really measure the return on investment in EM? Probably not, if we stay with “traditional” modeling
way and frameworks.
So, what would be the characteristics of efficient EM? We argue here that a multi-criteria approach
is needed to assess EM efficiency. First, it depends on the quality of the method and its components,
then the process of building and using models, including required expertise and responsibilities, has to
be considered, and, last but not least, tool support is vital for EM efficiency.
Khademhosseinieh and Seigerroth [11] propose a method, taking form of a set of criteria and
evaluation questions, for assessing the efficiency of EM methods. The criteria concern the quality of
the method product part (modeling perspective, components, concepts and their notation) and its
process part (framework, procedures, cooperation forms) but omit to evaluate the tool support. Also, it
is not possible to judge the effectiveness of the way of using the method.
Enterprise modeling, while certainly done using modeling tools and frameworks, remains a manual
task. It requires human resources having modeling skills and taking responsibility of building and
updating models. Therefore, the efficiency of EM heavily depends on the availability of modelers and
their expertise. In [12] Persson and Stirna stress that “much of the success of projects using EM depends
on how the EM process is organized and on the competence level of the expert responsible for the EM
approach”. They define a set of core EM competences classified in two categories: those related to
modeling itself (ability to model, to facilitate modeling sessions) and those related to setting up and
managing EM projects.
However, having one or two EM experts, especially if they are not necessarily aware of all
organizational changes, may not be enough. Given current business dynamics and interorganizational
scopes, cross-organizational and even inter-organizational collaboration in EM becomes vital.
Collaborative modeling itself is not a new topic, many works have been published (e.g. [13, 14]). They
highlight challenges, analyze related problems and propose solutions to collaborative modeling
sessions. Though, the motivation and responsibility issues, which are critical in EM, are not discussed.
Beyond the need of collaboration in EM, we should also consider the need of contribution from
various business actors as we are aiming to model increasingly complex business ecosystems.
Competencies in modeling does not necessary include domain expertise, but, both are necessary to build
consistent and precise enterprise models. Therefore, we advocate the need for a contributory EM
approach involving various stakeholders as co-creators of enterprise models, especially the “to-be”
ones. Involving stakeholders in EM is of course a challenge, but it is worth of considering. The
inspiration for building such an approach could be taken from [15] where a contributory approach is
proposed to the co-creation of transdisciplinary information service.
To be efficient, enterprise models need to be dynamic, i.e. to evolve on-the-fly. The “as-is” models
should always reflect the situation of the day not that of yesterday, which means they need to evolve at
any change in the organization. This is a real challenge as the current modeling frameworks and tools
do not support dynamic evolution of models. Quite often there is a lag between the evolution the
organization and its information systems and the updating of related enterprise models. Also, enterprise
models should help to simulate enterprise evolution – to get the picture of the “to-be” enterprise. But,
that requires going beyond the classic EM and its current tools. A framework for IS evolution steering
proposed in [16] goes towards this direction. It provides a kernel model and guidance for developing
67
an Informational Steering Information System (ISIS). ISIS is a meta-IS to drive organizational and IS
changes in an interdependent manner and thus guarantee their consistency. Even if this approach does
not support modeling in its traditional way, as no graphical representations are used, however, it allows
to simulate IS evolution and so enterprise evolution. The next step, would be to extend ISIS with
features allowing to generate enterprise models and thus facilitate timely deployment of IS solutions.
5. Conclusions
In this section, we summarize the key findings from the panel discussion as it was presented in sections
2-4.
Modeling frameworks need to ensure objectivity. The concepts of EM and EA frameworks, such
as goals, capabilities, services, processes, resources, with their underlying infrastructure - technologies
and hardware are, from the social perspective, fundamentally neutral; however, they may become
inherently biased based on the way we design and use them. Bias can emerge from improper sized or
structured data that instantiate these concepts, or for example from different policies of use of a
technical infrastructure in different business, demographic, or other types of contexts; some illustrative
examples are given in section 2. This relevant threat cannot be left unchecked, i.e. the modeling
frameworks need to reinforce systemic mechanisms for ensuring resilience of the objectivity of
instances of modeled entities. The mechanisms could be different - explicit requiring some modeling
extensions of the current frameworks; or rather implicit by enforcing additional constraints or quality
requirements in the modeling process of the key concepts.
Models need to be situational and embracing data. Enterprise modeling methods and languages
are, in general, designed for explicitly capturing and modeling data, as well as for development of
information processing solutions. However, increasingly emerging smart connected products and
context-aware business services are some examples that use huge bulks of data, and as such, they pose
the challenges for managing this data within the frameworks, as discussed in section 3. Beyond the high
volume and dynamics of the data, additional challenges relate to the complexity of the products’ and
services’ indicators due to their amounts and mutual dependency. A known mechanism of EM
techniques for handling context-awareness and variability in systems is adaptation – yet it is often
expensive in terms of modeling effort. When EM is applied in the situations where information flow
should be considered as complete and exact as possible, extending the frameworks by “plug-in” meta-
modeling components, dynamic extensions of modeling languages, or generation of models from event
or data streams, may be a right solution.
Modeling need to be efficient. Enterprise modeling is an overarching and effort activity; yet, once
produced, the models have a limited lifespan as they are the blueprints of a time-stamped situation.
Digital transformation even more increases this challenge owing to its high dynamic nature. This leads
to the questions for how to assess efficiency of EM. One logical answer leads to the multi-criteria
approach - the quality of the modeling method and its components, the process of building and using
models, as well as with a reliable tool support. In addition, enterprise modeling remains a human task
– it requires human resources having modeling competencies and taking responsibilities for building
and updating models, as well as those related to setting up and managing EM projects. Furthermore,
interorganizational scopes, make collaboration in EM to become vital, and where the motivation and
responsibility issues are of the essential importance. When modeling complex business ecosystems,
the needed competencies go beyond domain expertise; they require contributory EM approach
involving various stakeholders as co-creators of models.
Enterprise models need to be dynamic and right, i.e. to evolve on-the-fly in the correct direction.
The “as-is” models should always reflect the situation of the day not that of yesterday, which means
they need to evolve at any change in the organization. This raises the challenge for enterprise models
to continue being design time constructs and even more than today target runtime artefacts.
6. References
[1] WMP van der Aalst, J. Becker, M. Bichler, HU Buhl, J. Dibbern, U. Frank, U. Hasenkamp, A.
Heinzl, O. Hinz, K-L Hui, M. Jarke, D. Karagiannis, N. Kliewer, W. König, J. Mendling, P.
68
Mertens, M. Rossi, S. Voss, C. Weinhardt, R. Winter, J. Zdravkovic. Views on the Past, Present,
and Future of Business and Information Systems Engineering. Bus Inf Syst Eng 60 (6):443-477 60
(6):443-477 (2018)
[2] K. Sandkuhl, J. Stirna, A. Persson, M. Wißotzki, Enterprise modeling. Springer, Berlin, Heidelberg
(2014)
[3] C. H. Tsai, J. Zdravkovic, J. Stirna, Capability Management of Digital Business Ecosystems – A
Case of Resilience Modeling in the Healthcare Domain. In: Proceedings of the 32nd. International
Conference on Advanced Information Systems Engineering, volume 386 of CAiSE2020, LNBIP
Springer, Cham, pp. 126–137 (2020)
[4] F. Björck, M. Henkel, J. Stirna, J. Zdravkovic. Cyber Resilience – Fundamentals for a Definition.
In: Proceedings of the conf. New Contributions in Information Systems and Technologies.
Advances in Intelligent Systems and Computing, vol 353. Springer, Cham (2015)
[5] C. Breslin. Bias Creeps into Technology—We can’t avoid bringing our own perspective to the
products we build. Towards data science. [online] 2020 Available at:
https://towardsdatascience.com/bias-creeps-into-technology-e221dd7eae76
[6] R. Benjamin, R. Race after technology: Abolitionist tools for the new Jim code. Cambridge: Polity,
285 pp. (2019)
[7] J. Buolamwini, T. Gebru. Gender shades: Intersectional accuracy disparities in commercial gender
classification. In Proceedings of the 1st Conference on Fairness, Accountability and Transparency,
PMLR 81:77-91 (2018)
[8] D. van der Linden, O.A. Michalec, A. Zamansky. Cybersecurity for smart farming: socio-cultural
context matters. IEEE Technology and Society Magazine (2020)
[9] M. Porter, J Heppelmann. How smart, connected products are transforming competition. Harvard
Business Review, 92(11), 64-88 (2014)
[10] S. Bērziša, G. Bravos, T. Cardona Gonzalez, U. Czubayko, S. España, J. Grabis, M. Henkel, L.
Jokste, J. Kampars, H. Koç, J-C Kuhr, C. Llorca, P. Loucopoulos, R. J. Pascual, O. Pastor, K.
Sandkuhl, H. Simic, J. Stirna, F. G. Valverde, J. Zdravkovic. Capability driven development: an
approach to designing digital enterprises. Business & Information Systems Engineering, 57(1), 15-
25 (2015)
[11] B. Khademhosseinieh, U. Seigerroth. Towards Evaluating Efficiency of Enterprise Modeling
Methods. In: Skersys T., Butleris R., Butkiene R. (eds) Information and Software Technologies.
ICIST 2012. Communications in Computer and Information Science, vol 319. Springer, Berlin,
Heidelberg (2012)
[12] A. Persson, J. Stirna. Towards Defining a Competence Profile for the Enterprise Modeling
Practitioner. In: Proceedings of the Work. Conf. Practice of Enterprise Modeling, PoEM 2010,
LNBIP 68, pp. 232–245 (2010)
[13] P. Rittgen. Collaborative Modelling – A Design Science Approach. Proceedings of the 42nd Hawaii
International Conference on Systems Science, 2009, pp. 1– 10 (2009)
[14] P. Rittgen. Success Factors of e-Collaboration in Business Process Modeling. Proceedings of
CAiSE 2010, Springer LNCS 6051, pp.24–37 (2010)
[15] J. Ralyté, J., M. Léonard. Tiers-Lieu for Services: An Exploratory Approach to Societal
Progression. In: Proc. of the Conference Exploring Service Science. IESS 2020, LNBIP 377,
Springer, pp. 289-303 (2020)
[16] W. Opprecht, J. Ralyté, J., M. Léonard. Towards a Framework for Enterprise Information System
Evolution Steering. In: Proc. of The Practice of Enterprise Modeling, 7th IFIP WG 8.1 Working
Conference PoEM2014, LNBIP 197, pp. 118-132, Springer, Berlin, Heidelberg (2014)
69