Content uploaded by Benjamin Rolf
Author content
All content in this area was uploaded by Benjamin Rolf on Oct 29, 2022
Content may be subject to copyright.
IFAC PapersOnLine 55-10 (2022) 1693–1698
ScienceDirect
ScienceDirect
Available online at www.sciencedirect.com
2405-8963 Copyright © 2022 The Authors. This is an open access article under the CC BY-NC-ND license
.
Peer review under responsibility of International Federation of Automatic Control.
10.1016/j.ifacol.2022.09.641
10.1016/j.ifacol.2022.09.641 2405-8963
Copyright ©
2022 The Authors. This is an open access article under the CC BY-NC-ND license
(
https://creativecommons.org/licenses/by-nc-nd/4.0/
)
Using Knowledge Graphs and
Human-Centric Artificial Intelligence for
Reconfigurable Supply Chains: A Research
Framework
Benjamin Rolf ∗Nasser Mebarki ∗∗ Sebastian Lang ∗∗∗
Tobias Reggelin ∗Olivier Cardin ∗∗ Harold Mouch`ere ∗∗
Alexandre Dolgui ∗∗∗∗
∗Otto-von-Guericke-University, Magdeburg, 39106, Germany (e-mail:
benjamin.rolf@ovgu.de, tobias.reggelin@ovgu.de).
∗∗ Nantes University, LS2N UMR-CNRS 6004, Nantes, France
(e-mail: surname.lastname@univ-nantes.fr).
∗∗∗ Fraunhofer Institute for Factory Operation and Automation IFF,
Magdeburg, 39106, Germany (e-mail:
sebastian.lang@iff.fraunhofer.de).
∗∗∗∗ IMT-Atlantique, LS2N UMR-CNRS 6004, Nantes, France (e-mail:
alexandre.dolgui@imt-atlantique.fr).
Abstract: Reconfigurable supply chains received increasing interest from academia and
industry in the past years, especially because recent events such as the COVID-19 pandemic
revealed the vulnerability of present supply chains. Especially the rapid digitalization and the
emergence of artificial intelligence in supply chain management create new opportunities for
implementing reconfigurable supply chains. In this paper, we propose a concept that uses a
knowledge graph and graph-based artificial intelligence to recommend reconfigurations of the
supply chain. Due to the importance of these strategic decisions, the concept considers a human-
centric approach using a recommender system. The knowledge graph exploits the inherent graph
structure of supply chains to create a machine-readable and human-understandable database
that also supports reasoning. Finally, we provide three research hypotheses and examine the
integration of open-source knowledge graphs and news monitoring tools.
Keywords: Reconfigurable supply chains; Knowledge graphs; Supply chain management;
Human-centric artificial intelligence; Graph neural networks; Open data
1. INTRODUCTION
Supply chain configuration and operation are highly rele-
vant for corporate success and have been topics of interest
for many years in academia and industry. Many supply
chains became highly optimized structures focusing on
efficiency and profit because companies took globalization,
low tariffs, and stable trading rules for granted. However,
in the last years the global situation became increasingly
unstable, and several recent events revealed the vulnera-
bility of global supply chains. Especially the temporary
closures of factories in different regions around the world
caused by the COVID-19 pandemic affected supply chains.
In combination with other events such as the trade war
between the USA and China, the withdrawal of the UK
from the European Union or the temporary Suez Canal
obstruction, the disruptions became a serious threat that
lead to shortages of different products. Stricter entry regu-
lations resulted in longer and unpredictable lead times and
broke long-distance supply chains with low inventories.
Currently, the situation does not seem to be improving
soon, also due to more frequent extreme weather events
and new outbreaks of infectious diseases caused by the
climate change (Javorcik, 2020; Schroeder et al., 2021).
As a result of these trends, many companies started to
rethink their supply chain configuration and shift from
a pure focus on efficiency and profit to a more resilient
operation (Javorcik, 2020). A reaction to the COVID-19
pandemic would have required to quickly retake strategic
decisions such as supply chain design and supplier selec-
tion that usually have a planning horizon of 2-5 years.
Current supply chains were simply not designed for that.
The concept of reconfigurable supply chains that focus
on the adaption to changing circumstances by dynami-
cally adding, removing, or replacing units in the supply
chain offers a possible solution to decrease the reaction
time (Dolgui et al., 2020). A recent survey by Schroeder
et al. (2021) showed that only 20% of shipping companies
and logstics service providers had a comprehensive risk
management before the pandemic but the procurement of
64% was negatively affected by the pandemic. Further-
more, 70% of the companies plan to invest in business
analytics, artificial intelligence (AI), and the digitalization
of business processes after the pandemic which creates
an urgent need for further research on how to improve
risk management and decision-making in supply chains.
Using Knowledge Graphs and
Human-Centric Artificial Intelligence for
Reconfigurable Supply Chains: A Research
Framework
Benjamin Rolf ∗Nasser Mebarki ∗∗ Sebastian Lang ∗∗∗
Tobias Reggelin ∗Olivier Cardin ∗∗ Harold Mouch`ere ∗∗
Alexandre Dolgui ∗∗∗∗
∗Otto-von-Guericke-University, Magdeburg, 39106, Germany (e-mail:
benjamin.rolf@ovgu.de, tobias.reggelin@ovgu.de).
∗∗ Nantes University, LS2N UMR-CNRS 6004, Nantes, France
(e-mail: surname.lastname@univ-nantes.fr).
∗∗∗ Fraunhofer Institute for Factory Operation and Automation IFF,
Magdeburg, 39106, Germany (e-mail:
sebastian.lang@iff.fraunhofer.de).
∗∗∗∗ IMT-Atlantique, LS2N UMR-CNRS 6004, Nantes, France (e-mail:
alexandre.dolgui@imt-atlantique.fr).
Abstract: Reconfigurable supply chains received increasing interest from academia and
industry in the past years, especially because recent events such as the COVID-19 pandemic
revealed the vulnerability of present supply chains. Especially the rapid digitalization and the
emergence of artificial intelligence in supply chain management create new opportunities for
implementing reconfigurable supply chains. In this paper, we propose a concept that uses a
knowledge graph and graph-based artificial intelligence to recommend reconfigurations of the
supply chain. Due to the importance of these strategic decisions, the concept considers a human-
centric approach using a recommender system. The knowledge graph exploits the inherent graph
structure of supply chains to create a machine-readable and human-understandable database
that also supports reasoning. Finally, we provide three research hypotheses and examine the
integration of open-source knowledge graphs and news monitoring tools.
Keywords: Reconfigurable supply chains; Knowledge graphs; Supply chain management;
Human-centric artificial intelligence; Graph neural networks; Open data
1. INTRODUCTION
Supply chain configuration and operation are highly rele-
vant for corporate success and have been topics of interest
for many years in academia and industry. Many supply
chains became highly optimized structures focusing on
efficiency and profit because companies took globalization,
low tariffs, and stable trading rules for granted. However,
in the last years the global situation became increasingly
unstable, and several recent events revealed the vulnera-
bility of global supply chains. Especially the temporary
closures of factories in different regions around the world
caused by the COVID-19 pandemic affected supply chains.
In combination with other events such as the trade war
between the USA and China, the withdrawal of the UK
from the European Union or the temporary Suez Canal
obstruction, the disruptions became a serious threat that
lead to shortages of different products. Stricter entry regu-
lations resulted in longer and unpredictable lead times and
broke long-distance supply chains with low inventories.
Currently, the situation does not seem to be improving
soon, also due to more frequent extreme weather events
and new outbreaks of infectious diseases caused by the
climate change (Javorcik, 2020; Schroeder et al., 2021).
As a result of these trends, many companies started to
rethink their supply chain configuration and shift from
a pure focus on efficiency and profit to a more resilient
operation (Javorcik, 2020). A reaction to the COVID-19
pandemic would have required to quickly retake strategic
decisions such as supply chain design and supplier selec-
tion that usually have a planning horizon of 2-5 years.
Current supply chains were simply not designed for that.
The concept of reconfigurable supply chains that focus
on the adaption to changing circumstances by dynami-
cally adding, removing, or replacing units in the supply
chain offers a possible solution to decrease the reaction
time (Dolgui et al., 2020). A recent survey by Schroeder
et al. (2021) showed that only 20% of shipping companies
and logstics service providers had a comprehensive risk
management before the pandemic but the procurement of
64% was negatively affected by the pandemic. Further-
more, 70% of the companies plan to invest in business
analytics, artificial intelligence (AI), and the digitalization
of business processes after the pandemic which creates
an urgent need for further research on how to improve
risk management and decision-making in supply chains.
Using Knowledge Graphs and
Human-Centric Artificial Intelligence for
Reconfigurable Supply Chains: A Research
Framework
Benjamin Rolf ∗Nasser Mebarki ∗∗ Sebastian Lang ∗∗∗
Tobias Reggelin ∗Olivier Cardin ∗∗ Harold Mouch`ere ∗∗
Alexandre Dolgui ∗∗∗∗
∗Otto-von-Guericke-University, Magdeburg, 39106, Germany (e-mail:
benjamin.rolf@ovgu.de, tobias.reggelin@ovgu.de).
∗∗ Nantes University, LS2N UMR-CNRS 6004, Nantes, France
(e-mail: surname.lastname@univ-nantes.fr).
∗∗∗ Fraunhofer Institute for Factory Operation and Automation IFF,
Magdeburg, 39106, Germany (e-mail:
sebastian.lang@iff.fraunhofer.de).
∗∗∗∗ IMT-Atlantique, LS2N UMR-CNRS 6004, Nantes, France (e-mail:
alexandre.dolgui@imt-atlantique.fr).
Abstract: Reconfigurable supply chains received increasing interest from academia and
industry in the past years, especially because recent events such as the COVID-19 pandemic
revealed the vulnerability of present supply chains. Especially the rapid digitalization and the
emergence of artificial intelligence in supply chain management create new opportunities for
implementing reconfigurable supply chains. In this paper, we propose a concept that uses a
knowledge graph and graph-based artificial intelligence to recommend reconfigurations of the
supply chain. Due to the importance of these strategic decisions, the concept considers a human-
centric approach using a recommender system. The knowledge graph exploits the inherent graph
structure of supply chains to create a machine-readable and human-understandable database
that also supports reasoning. Finally, we provide three research hypotheses and examine the
integration of open-source knowledge graphs and news monitoring tools.
Keywords: Reconfigurable supply chains; Knowledge graphs; Supply chain management;
Human-centric artificial intelligence; Graph neural networks; Open data
1. INTRODUCTION
Supply chain configuration and operation are highly rele-
vant for corporate success and have been topics of interest
for many years in academia and industry. Many supply
chains became highly optimized structures focusing on
efficiency and profit because companies took globalization,
low tariffs, and stable trading rules for granted. However,
in the last years the global situation became increasingly
unstable, and several recent events revealed the vulnera-
bility of global supply chains. Especially the temporary
closures of factories in different regions around the world
caused by the COVID-19 pandemic affected supply chains.
In combination with other events such as the trade war
between the USA and China, the withdrawal of the UK
from the European Union or the temporary Suez Canal
obstruction, the disruptions became a serious threat that
lead to shortages of different products. Stricter entry regu-
lations resulted in longer and unpredictable lead times and
broke long-distance supply chains with low inventories.
Currently, the situation does not seem to be improving
soon, also due to more frequent extreme weather events
and new outbreaks of infectious diseases caused by the
climate change (Javorcik, 2020; Schroeder et al., 2021).
As a result of these trends, many companies started to
rethink their supply chain configuration and shift from
a pure focus on efficiency and profit to a more resilient
operation (Javorcik, 2020). A reaction to the COVID-19
pandemic would have required to quickly retake strategic
decisions such as supply chain design and supplier selec-
tion that usually have a planning horizon of 2-5 years.
Current supply chains were simply not designed for that.
The concept of reconfigurable supply chains that focus
on the adaption to changing circumstances by dynami-
cally adding, removing, or replacing units in the supply
chain offers a possible solution to decrease the reaction
time (Dolgui et al., 2020). A recent survey by Schroeder
et al. (2021) showed that only 20% of shipping companies
and logstics service providers had a comprehensive risk
management before the pandemic but the procurement of
64% was negatively affected by the pandemic. Further-
more, 70% of the companies plan to invest in business
analytics, artificial intelligence (AI), and the digitalization
of business processes after the pandemic which creates
an urgent need for further research on how to improve
risk management and decision-making in supply chains.
Using Knowledge Graphs and
Human-Centric Artificial Intelligence for
Reconfigurable Supply Chains: A Research
Framework
Benjamin Rolf ∗Nasser Mebarki ∗∗ Sebastian Lang ∗∗∗
Tobias Reggelin ∗Olivier Cardin ∗∗ Harold Mouch`ere ∗∗
Alexandre Dolgui ∗∗∗∗
∗
Otto-von-Guericke-University, Magdeburg, 39106, Germany (e-mail:
benjamin.rolf@ovgu.de, tobias.reggelin@ovgu.de).
∗∗ Nantes University, LS2N UMR-CNRS 6004, Nantes, France
(e-mail: surname.lastname@univ-nantes.fr).
∗∗∗ Fraunhofer Institute for Factory Operation and Automation IFF,
Magdeburg, 39106, Germany (e-mail:
sebastian.lang@iff.fraunhofer.de).
∗∗∗∗ IMT-Atlantique, LS2N UMR-CNRS 6004, Nantes, France (e-mail:
alexandre.dolgui@imt-atlantique.fr).
Abstract: Reconfigurable supply chains received increasing interest from academia and
industry in the past years, especially because recent events such as the COVID-19 pandemic
revealed the vulnerability of present supply chains. Especially the rapid digitalization and the
emergence of artificial intelligence in supply chain management create new opportunities for
implementing reconfigurable supply chains. In this paper, we propose a concept that uses a
knowledge graph and graph-based artificial intelligence to recommend reconfigurations of the
supply chain. Due to the importance of these strategic decisions, the concept considers a human-
centric approach using a recommender system. The knowledge graph exploits the inherent graph
structure of supply chains to create a machine-readable and human-understandable database
that also supports reasoning. Finally, we provide three research hypotheses and examine the
integration of open-source knowledge graphs and news monitoring tools.
Keywords: Reconfigurable supply chains; Knowledge graphs; Supply chain management;
Human-centric artificial intelligence; Graph neural networks; Open data
1. INTRODUCTION
Supply chain configuration and operation are highly rele-
vant for corporate success and have been topics of interest
for many years in academia and industry. Many supply
chains became highly optimized structures focusing on
efficiency and profit because companies took globalization,
low tariffs, and stable trading rules for granted. However,
in the last years the global situation became increasingly
unstable, and several recent events revealed the vulnera-
bility of global supply chains. Especially the temporary
closures of factories in different regions around the world
caused by the COVID-19 pandemic affected supply chains.
In combination with other events such as the trade war
between the USA and China, the withdrawal of the UK
from the European Union or the temporary Suez Canal
obstruction, the disruptions became a serious threat that
lead to shortages of different products. Stricter entry regu-
lations resulted in longer and unpredictable lead times and
broke long-distance supply chains with low inventories.
Currently, the situation does not seem to be improving
soon, also due to more frequent extreme weather events
and new outbreaks of infectious diseases caused by the
climate change (Javorcik, 2020; Schroeder et al., 2021).
As a result of these trends, many companies started to
rethink their supply chain configuration and shift from
a pure focus on efficiency and profit to a more resilient
operation (Javorcik, 2020). A reaction to the COVID-19
pandemic would have required to quickly retake strategic
decisions such as supply chain design and supplier selec-
tion that usually have a planning horizon of 2-5 years.
Current supply chains were simply not designed for that.
The concept of reconfigurable supply chains that focus
on the adaption to changing circumstances by dynami-
cally adding, removing, or replacing units in the supply
chain offers a possible solution to decrease the reaction
time (Dolgui et al., 2020). A recent survey by Schroeder
et al. (2021) showed that only 20% of shipping companies
and logstics service providers had a comprehensive risk
management before the pandemic but the procurement of
64% was negatively affected by the pandemic. Further-
more, 70% of the companies plan to invest in business
analytics, artificial intelligence (AI), and the digitalization
of business processes after the pandemic which creates
an urgent need for further research on how to improve
risk management and decision-making in supply chains.
Using Knowledge Graphs and
Human-Centric Artificial Intelligence for
Reconfigurable Supply Chains: A Research
Framework
Benjamin Rolf ∗Nasser Mebarki ∗∗ Sebastian Lang ∗∗∗
Tobias Reggelin ∗Olivier Cardin ∗∗ Harold Mouch`ere ∗∗
Alexandre Dolgui
∗∗∗∗
∗Otto-von-Guericke-University, Magdeburg, 39106, Germany (e-mail:
benjamin.rolf@ovgu.de, tobias.reggelin@ovgu.de).
∗∗ Nantes University, LS2N UMR-CNRS 6004, Nantes, France
(e-mail: surname.lastname@univ-nantes.fr).
∗∗∗ Fraunhofer Institute for Factory Operation and Automation IFF,
Magdeburg, 39106, Germany (e-mail:
sebastian.lang@iff.fraunhofer.de).
∗∗∗∗ IMT-Atlantique, LS2N UMR-CNRS 6004, Nantes, France (e-mail:
alexandre.dolgui@imt-atlantique.fr).
Abstract: Reconfigurable supply chains received increasing interest from academia and
industry in the past years, especially because recent events such as the COVID-19 pandemic
revealed the vulnerability of present supply chains. Especially the rapid digitalization and the
emergence of artificial intelligence in supply chain management create new opportunities for
implementing reconfigurable supply chains. In this paper, we propose a concept that uses a
knowledge graph and graph-based artificial intelligence to recommend reconfigurations of the
supply chain. Due to the importance of these strategic decisions, the concept considers a human-
centric approach using a recommender system. The knowledge graph exploits the inherent graph
structure of supply chains to create a machine-readable and human-understandable database
that also supports reasoning. Finally, we provide three research hypotheses and examine the
integration of open-source knowledge graphs and news monitoring tools.
Keywords: Reconfigurable supply chains; Knowledge graphs; Supply chain management;
Human-centric artificial intelligence; Graph neural networks; Open data
1. INTRODUCTION
Supply chain configuration and operation are highly rele-
vant for corporate success and have been topics of interest
for many years in academia and industry. Many supply
chains became highly optimized structures focusing on
efficiency and profit because companies took globalization,
low tariffs, and stable trading rules for granted. However,
in the last years the global situation became increasingly
unstable, and several recent events revealed the vulnera-
bility of global supply chains. Especially the temporary
closures of factories in different regions around the world
caused by the COVID-19 pandemic affected supply chains.
In combination with other events such as the trade war
between the USA and China, the withdrawal of the UK
from the European Union or the temporary Suez Canal
obstruction, the disruptions became a serious threat that
lead to shortages of different products. Stricter entry regu-
lations resulted in longer and unpredictable lead times and
broke long-distance supply chains with low inventories.
Currently, the situation does not seem to be improving
soon, also due to more frequent extreme weather events
and new outbreaks of infectious diseases caused by the
climate change (Javorcik, 2020; Schroeder et al., 2021).
As a result of these trends, many companies started to
rethink their supply chain configuration and shift from
a pure focus on efficiency and profit to a more resilient
operation (Javorcik, 2020). A reaction to the COVID-19
pandemic would have required to quickly retake strategic
decisions such as supply chain design and supplier selec-
tion that usually have a planning horizon of 2-5 years.
Current supply chains were simply not designed for that.
The concept of reconfigurable supply chains that focus
on the adaption to changing circumstances by dynami-
cally adding, removing, or replacing units in the supply
chain offers a possible solution to decrease the reaction
time (Dolgui et al., 2020). A recent survey by Schroeder
et al. (2021) showed that only 20% of shipping companies
and logstics service providers had a comprehensive risk
management before the pandemic but the procurement of
64% was negatively affected by the pandemic. Further-
more, 70% of the companies plan to invest in business
analytics, artificial intelligence (AI), and the digitalization
of business processes after the pandemic which creates
an urgent need for further research on how to improve
risk management and decision-making in supply chains.
Using Knowledge Graphs and
Human-Centric Artificial Intelligence for
Reconfigurable Supply Chains: A Research
Framework
Benjamin Rolf ∗Nasser Mebarki ∗∗ Sebastian Lang ∗∗∗
Tobias Reggelin ∗Olivier Cardin ∗∗ Harold Mouch`ere ∗∗
Alexandre Dolgui ∗∗∗∗
∗Otto-von-Guericke-University, Magdeburg, 39106, Germany (e-mail:
benjamin.rolf@ovgu.de, tobias.reggelin@ovgu.de).
∗∗ Nantes University, LS2N UMR-CNRS 6004, Nantes, France
(e-mail: surname.lastname@univ-nantes.fr).
∗∗∗ Fraunhofer Institute for Factory Operation and Automation IFF,
Magdeburg, 39106, Germany (e-mail:
sebastian.lang@iff.fraunhofer.de).
∗∗∗∗ IMT-Atlantique, LS2N UMR-CNRS 6004, Nantes, France (e-mail:
alexandre.dolgui@imt-atlantique.fr).
Abstract: Reconfigurable supply chains received increasing interest from academia and
industry in the past years, especially because recent events such as the COVID-19 pandemic
revealed the vulnerability of present supply chains. Especially the rapid digitalization and the
emergence of artificial intelligence in supply chain management create new opportunities for
implementing reconfigurable supply chains. In this paper, we propose a concept that uses a
knowledge graph and graph-based artificial intelligence to recommend reconfigurations of the
supply chain. Due to the importance of these strategic decisions, the concept considers a human-
centric approach using a recommender system. The knowledge graph exploits the inherent graph
structure of supply chains to create a machine-readable and human-understandable database
that also supports reasoning. Finally, we provide three research hypotheses and examine the
integration of open-source knowledge graphs and news monitoring tools.
Keywords: Reconfigurable supply chains; Knowledge graphs; Supply chain management;
Human-centric artificial intelligence; Graph neural networks; Open data
1. INTRODUCTION
Supply chain configuration and operation are highly rele-
vant for corporate success and have been topics of interest
for many years in academia and industry. Many supply
chains became highly optimized structures focusing on
efficiency and profit because companies took globalization,
low tariffs, and stable trading rules for granted. However,
in the last years the global situation became increasingly
unstable, and several recent events revealed the vulnera-
bility of global supply chains. Especially the temporary
closures of factories in different regions around the world
caused by the COVID-19 pandemic affected supply chains.
In combination with other events such as the trade war
between the USA and China, the withdrawal of the UK
from the European Union or the temporary Suez Canal
obstruction, the disruptions became a serious threat that
lead to shortages of different products. Stricter entry regu-
lations resulted in longer and unpredictable lead times and
broke long-distance supply chains with low inventories.
Currently, the situation does not seem to be improving
soon, also due to more frequent extreme weather events
and new outbreaks of infectious diseases caused by the
climate change (Javorcik, 2020; Schroeder et al., 2021).
As a result of these trends, many companies started to
rethink their supply chain configuration and shift from
a pure focus on efficiency and profit to a more resilient
operation (Javorcik, 2020). A reaction to the COVID-19
pandemic would have required to quickly retake strategic
decisions such as supply chain design and supplier selec-
tion that usually have a planning horizon of 2-5 years.
Current supply chains were simply not designed for that.
The concept of reconfigurable supply chains that focus
on the adaption to changing circumstances by dynami-
cally adding, removing, or replacing units in the supply
chain offers a possible solution to decrease the reaction
time (Dolgui et al., 2020). A recent survey by Schroeder
et al. (2021) showed that only 20% of shipping companies
and logstics service providers had a comprehensive risk
management before the pandemic but the procurement of
64% was negatively affected by the pandemic. Further-
more, 70% of the companies plan to invest in business
analytics, artificial intelligence (AI), and the digitalization
of business processes after the pandemic which creates
an urgent need for further research on how to improve
risk management and decision-making in supply chains.
1694 Benjamin Rolf et al. / IFAC PapersOnLine 55-10 (2022) 1693–1698
However, literature on reconfigurable supply chains is still
limited to highly conceptual considerations that all agree
that business analytics and AI are important enablers
(Chandra and Grabis, 2016; Dolgui et al., 2020; Kelepouris
et al., 2006). Therefore, we identify research gaps and
propose a research framework for enabling reconfigurable
supply chains with human-centric AI in this paper. Section
2 summarizes the existing literature on reconfigurable sup-
ply chains and reconfigurable manufacturing systems. Sec-
tion 3 presents the state-of-the-art on knowledge graphs
in supply chain management because knowledge graphs
are an important basis for the research framework that
is presented in section 4 and section 5. Finally, we draw
conclusions on future research in section 6.
2. RECONFIGURABLE SUPPLY CHAINS
To present the current state-of-the-art for reconfigurable
supply chains it is important to give a short overview on re-
configurable manufacturing systems first. Reconfigurable
manufacturing systems were first proposed by Koren et al.
(1999) as a reaction to unpredictable market changes
that occurred with increasing pace in the late 1990s.
These challenges include for example decreasing product
life cycles, fluctuations in demand and product mix, new
government regulations regarding safety and environment,
and fast changing process technology. The basic idea of
reconfigurable manufacturing systems is combining the
high throughput of dedicated production lines and the
flexibility of flexible manufacturing systems to build a
superior system that can react rapidly and cost-effectively
to changes. Other major drivers for introducing reconfig-
urable manufacturing systems were the rising globalization
and the increasingly global competition which require a
more responsive manufacturing system. Bortolini et al.
(2018) identified five main research streams in a recent lit-
erature review. The literature review showed that applied
research and field application constitute the most impor-
tant research stream which highlights the relevance not
only for academia but especially for industry. In addition,
current trends such as Industry 4.0 and cyber-physical
systems contributed to the dissemination of reconfigurable
manufacturing systems, and their key technologies enabled
the implementation in industrial environments (Qin et al.,
2016). Since the initial publication, the market challenges
have even intensified further (Koren et al., 2018). The
ongoing intensification of market challenges steadily shifts
the scope of reconfigurability from manufacturing systems
to entire supply chains which also caused the development
of a framework for reconfigurable supply chains. However,
a distinction must be made between reconfigurable supply
chains and classic supply chain reconfiguration. Classic
supply chain reconfiguration has been a topic of interest
since the late 1990s, but it was usually considered as a
one-time adaption of existing supply chains to new cir-
cumstances. On the contrary, reconfigurable supply chains
focus on the dynamic adaption to changing circumstances.
The concept received decent attention in the last few years
but does not have a lot of contributions yet. Reconfig-
urable supply chains were first proposed by Kelepouris
et al. (2006). Later, Chandra and Grabis (2016) as well
as Dolgui et al. (2020) published the most recognized
and comprehensive frameworks. Kelepouris et al. (2006)
define reconfigurable supply chains as networks that are
“designed for easy rearrangement or change of supply
network entities in a timely and cost-effective manner”.
The rearrangement of supply chain entities incorporates
the long-term decisions network design, supplier selection
and product program which usually have a long planning
horizon, involve high costs and significantly impact the
supply chain performance. Reconfigurable supply chains
describe the target concept where reconfigurations happen
dynamically as a consequence of certain events within a
short planning horizon (Chandra and Grabis, 2016; Zidi
et al., 2021a). Reconfigurability has six characteristics that
apply to reconfigurable manufacturing systems as well as
reconfigurable supply chains (Biswas et al., 2019; Kele-
pouris et al., 2006; Koren et al., 1999; Zidi et al., 2021a):
(1) Modularity enables the clustering into independent
modules.
(2) Integrability enables the introduction of new re-
sources, products, and processes.
(3) Convertibility enables the changeover between exist-
ing products.
(4) Diagnoseability enables the identification of problems
in the supply chain.
(5) Scalability enables the change of capacity.
(6) Customization enables the fast adaption within a
product family.
Reconfigurability has the potential to increase the robust-
ness, flexibility, and agility of supply chains. The dynamic
reconfiguration makes supply chains more robust to with-
stand disruptions such as natural disasters, labour disputes
or delivery problems. Not only disruptions can cause recon-
figurations but also delivery terms such as cheaper prices,
higher quality or shorter delivery times which leads to a
general flexibility and agility. However, reconfigurability
also raises new challenges. It further complicates supply
chain management by introducing a new decision-making
domain. Moreover, the fast change of suppliers hampers
supply chain collaboration and reduces the trust, impairs
information sharing and requires a high level of technology
in the manufacturing system for processing parts from
multiple suppliers (Chandra and Grabis, 2016). Dolgui
et al. (2020) consider reconfigurability as the central fac-
tor for achieving a digital, resilient, sustainable, and effi-
cient supply chain and highlight the relationships between
these requirements in a framework called X-Network. The
framework comprises a matrix with three levels and four
design principles that specifies the respective enabling
technologies. Furthermore, they introduce two new princi-
ples which are supply chain meta-structures and dynamic
autonomous services. According to them, a dynamic meta-
structure is an integrity of the organizational, product,
process-functional, information and financial structure of
a supply chain. A dynamic autonomous service is a compo-
sition of supply chain units within the meta-structure. For
example, a supply chain unit, respectively an enterprise,
can change its role in the process-functional structure and
shift from supplier to manufacturer. Chandra and Grabis
(2016) mention the three main aspects decision-making,
physical implementation, and logical implementation that
must be considered when implementing reconfigurable
supply chains. Decision-making determines when and how
to reconfigure supply chains. The physical implementation
comprises all structural requirements regarding facilities
and products that are necessary to enable reconfigurable
supply chains. The logical implementation considers the
design of business processes and information systems. Re-
cent technologies such as software, sensors and communi-
cation technologies can already meet the technological re-
quirements for reconfigurable supply chains. For instance,
Cˆandido et al. (2009) propose a service-oriented archi-
tecture for industrial automation at the shopfloor level.
Decision-making and logical implementation received less
attention. Instead, most contributions present theoretical
frameworks (Chandra and Grabis, 2016; Dolgui et al.,
2020; Kelepouris et al., 2006) or address special issues such
as modularity metrics (Zidi et al., 2021b), reconfigurability
metrics (Biswas et al., 2019) or performance evaluation
(Zidi et al., 2021c).
3. KNOWLEDGE GRAPHS IN SUPPLY CHAIN
MANAGEMENT
A knowledge graph is a simple but powerful data model
whose basic units are entities with attributes and relations.
They became popular in 2012 when Google complemented
its web search engine with knowledge graphs (Singhal,
2012). The terms knowledge graph and ontology have often
been used as synonyms, but according to the definition
review of Ehrlinger and W¨oß (2016), a knowledge graph
is superior and more complex because it also includes
a reasoning engine and integrates multiple data sources.
Reasoning engines for knowledge graphs are often based
on AI, for instance graph neural networks that are a
new class of machine learning algorithms for handling
graph structures as inputs (Arora, 2020). The promotion
of knowledge graphs was mainly driven by the semantic
web community as an effort to make the internet machine-
readable Ehrlinger and W¨oß (2016). However, three main
characteristics make knowledge graphs also interesting for
application in enterprises (Gomez-Perez et al., 2017; Zou,
2020):
(1) Knowledge graphs are machine-readable.
(2) Knowledge graphs are understandable for humans.
(3) Knowledge graphs extend information with context
and enable reasoning.
Knowledge graphs have been successfully applied in several
domains such as healthcare, cybersecurity, finance and
education (Zou, 2020). However, in supply chain manage-
ment, the application of knowledge graphs is still rare,
although supply chains face similar challenges as the other
domains: Increasingly available structured and unstruc-
tured data from different sources, rapid digitalization and
the emergence of AI (Schniederjans et al., 2020). Cur-
rently, many companies store their data in multiple iso-
lated data silos which indeed works for decision-making in
delimited domains. However, connecting all these data has
the potential to boost the performance of data analytics
and AI in supply chains (Schroeder et al., 2021; Song et al.,
2017). Knowledge graphs are a promising tool to create a
consistent database for analytics.
Especially the natural graph structure of supply chains
makes it suitable for the representation in form of graphs.
The few existing publications on knowledge graphs in sup-
ply chain management mainly focus on risk-related topics
and resilience and all of them were published in the last
few years. For instance, Zhang et al. (2019) implemented a
knowledge graph for supply chain credit risk assessment to
evaluate the risk of bankruptcy of suppliers. Ramzy et al.
(2021) propose a dynamic pricing model for semiconductor
supply chains based on a knowledge graph. Another use
case is the application of link prediction algorithms that
have their origin in biological networks and social network
analytics (L¨u and Zhou, 2011). Link prediction has the
objective to reveal hidden connections in large supply
networks that are unknown from the point of view of
a single company. By this, companies can increase their
knowledge on unknown lower-tier or higher-tier suppliers.
Brintrup et al. (2018) use na¨ıve Bayes and logistic regres-
sion to estimate the likelihood that two suppliers interact
with each other based on the supply network structure.
Aziz et al. (2021) and Kosasih and Brintrup (2021) use
graph neural networks for the same task and achieve
an accuracy of more than 85%. These three approaches
are promising examples of the deployment of knowledge
graphs in supply chain management. Ontologies are more
widespread in supply chain research, but their purpose is
slightly different from knowledge graphs. They are usually
used for modelling supply chain processes and information
systems in a rather abstract manner. For instance, Ameri
and Kulvatunyou (2019) develop a supply chain reference
ontology to make supply chains interoperable and improve
collaboration. Geerts and O’Leary (2014) describe their
ontology as a supply chain of things that has the objective
to gather information on the location of objects along the
supply chain. Grubic and Fan (2010) present a literature
review on supply chain ontologies.
4. RESEARCH NEED AND RESEARCH
FRAMEWORK
As the reasoning engines of knowledge graphs are designed
to adjust the edges and weights, use cases in supply chain
management where the relations are dynamic are most
interesting. So far, there were only very few approaches
that exploit this characteristic because supply chains were
considered to be static. The only exception are the supply
chain visibility concepts of Brintrup et al. (2018), Aziz
et al. (2021) and Kosasih and Brintrup (2021). Recon-
figurable supply chains reveal a completely new use case
for knowledge graphs because the weights can be used to
model the dynamics of reconfigurations. For instance, the
weighted arcs can represent the probabilities of establish-
ing the respective supply relation. Regarding the previous
sections, we can identify two major research needs to be
addressed:
(1) A quantitative proof of concept for reconfig-
urable supply chains: Authors agree that recon-
figurable supply chains can potentially increase the
resilience, flexibility, and agility of supply chains but
there is no proof of concept. Furthermore, no bench-
mark problems or industrial case studies exist which
evaluate the performance of reconfigurable supply
chains in a quantitative way. However, companies
are ready to invest in technologies that increase the
resilience (Schroeder et al., 2021).
(2) AI solutions for supply chain management
based on knowledge graphs: Although character-
istics of knowledge graphs are promising to handle big
Benjamin Rolf et al. / IFAC PapersOnLine 55-10 (2022) 1693–1698 1695
and products that are necessary to enable reconfigurable
supply chains. The logical implementation considers the
design of business processes and information systems. Re-
cent technologies such as software, sensors and communi-
cation technologies can already meet the technological re-
quirements for reconfigurable supply chains. For instance,
Cˆandido et al. (2009) propose a service-oriented archi-
tecture for industrial automation at the shopfloor level.
Decision-making and logical implementation received less
attention. Instead, most contributions present theoretical
frameworks (Chandra and Grabis, 2016; Dolgui et al.,
2020; Kelepouris et al., 2006) or address special issues such
as modularity metrics (Zidi et al., 2021b), reconfigurability
metrics (Biswas et al., 2019) or performance evaluation
(Zidi et al., 2021c).
3. KNOWLEDGE GRAPHS IN SUPPLY CHAIN
MANAGEMENT
A knowledge graph is a simple but powerful data model
whose basic units are entities with attributes and relations.
They became popular in 2012 when Google complemented
its web search engine with knowledge graphs (Singhal,
2012). The terms knowledge graph and ontology have often
been used as synonyms, but according to the definition
review of Ehrlinger and W¨oß (2016), a knowledge graph
is superior and more complex because it also includes
a reasoning engine and integrates multiple data sources.
Reasoning engines for knowledge graphs are often based
on AI, for instance graph neural networks that are a
new class of machine learning algorithms for handling
graph structures as inputs (Arora, 2020). The promotion
of knowledge graphs was mainly driven by the semantic
web community as an effort to make the internet machine-
readable Ehrlinger and W¨oß (2016). However, three main
characteristics make knowledge graphs also interesting for
application in enterprises (Gomez-Perez et al., 2017; Zou,
2020):
(1) Knowledge graphs are machine-readable.
(2) Knowledge graphs are understandable for humans.
(3) Knowledge graphs extend information with context
and enable reasoning.
Knowledge graphs have been successfully applied in several
domains such as healthcare, cybersecurity, finance and
education (Zou, 2020). However, in supply chain manage-
ment, the application of knowledge graphs is still rare,
although supply chains face similar challenges as the other
domains: Increasingly available structured and unstruc-
tured data from different sources, rapid digitalization and
the emergence of AI (Schniederjans et al., 2020). Cur-
rently, many companies store their data in multiple iso-
lated data silos which indeed works for decision-making in
delimited domains. However, connecting all these data has
the potential to boost the performance of data analytics
and AI in supply chains (Schroeder et al., 2021; Song et al.,
2017). Knowledge graphs are a promising tool to create a
consistent database for analytics.
Especially the natural graph structure of supply chains
makes it suitable for the representation in form of graphs.
The few existing publications on knowledge graphs in sup-
ply chain management mainly focus on risk-related topics
and resilience and all of them were published in the last
few years. For instance, Zhang et al. (2019) implemented a
knowledge graph for supply chain credit risk assessment to
evaluate the risk of bankruptcy of suppliers. Ramzy et al.
(2021) propose a dynamic pricing model for semiconductor
supply chains based on a knowledge graph. Another use
case is the application of link prediction algorithms that
have their origin in biological networks and social network
analytics (L¨u and Zhou, 2011). Link prediction has the
objective to reveal hidden connections in large supply
networks that are unknown from the point of view of
a single company. By this, companies can increase their
knowledge on unknown lower-tier or higher-tier suppliers.
Brintrup et al. (2018) use na¨ıve Bayes and logistic regres-
sion to estimate the likelihood that two suppliers interact
with each other based on the supply network structure.
Aziz et al. (2021) and Kosasih and Brintrup (2021) use
graph neural networks for the same task and achieve
an accuracy of more than 85%. These three approaches
are promising examples of the deployment of knowledge
graphs in supply chain management. Ontologies are more
widespread in supply chain research, but their purpose is
slightly different from knowledge graphs. They are usually
used for modelling supply chain processes and information
systems in a rather abstract manner. For instance, Ameri
and Kulvatunyou (2019) develop a supply chain reference
ontology to make supply chains interoperable and improve
collaboration. Geerts and O’Leary (2014) describe their
ontology as a supply chain of things that has the objective
to gather information on the location of objects along the
supply chain. Grubic and Fan (2010) present a literature
review on supply chain ontologies.
4. RESEARCH NEED AND RESEARCH
FRAMEWORK
As the reasoning engines of knowledge graphs are designed
to adjust the edges and weights, use cases in supply chain
management where the relations are dynamic are most
interesting. So far, there were only very few approaches
that exploit this characteristic because supply chains were
considered to be static. The only exception are the supply
chain visibility concepts of Brintrup et al. (2018), Aziz
et al. (2021) and Kosasih and Brintrup (2021). Recon-
figurable supply chains reveal a completely new use case
for knowledge graphs because the weights can be used to
model the dynamics of reconfigurations. For instance, the
weighted arcs can represent the probabilities of establish-
ing the respective supply relation. Regarding the previous
sections, we can identify two major research needs to be
addressed:
(1) A quantitative proof of concept for reconfig-
urable supply chains: Authors agree that recon-
figurable supply chains can potentially increase the
resilience, flexibility, and agility of supply chains but
there is no proof of concept. Furthermore, no bench-
mark problems or industrial case studies exist which
evaluate the performance of reconfigurable supply
chains in a quantitative way. However, companies
are ready to invest in technologies that increase the
resilience (Schroeder et al., 2021).
(2) AI solutions for supply chain management
based on knowledge graphs: Although character-
istics of knowledge graphs are promising to handle big
1696 Benjamin Rolf et al. / IFAC PapersOnLine 55-10 (2022) 1693–1698
data in supply chain management, no case studies
exist that investigate the suitability of knowledge
graphs as a basis for AI algorithms in supply chain
management.
We think that a quantitative proof of concept for reconfig-
urable supply chains with a focus on knowledge manage-
ment and human-centric AI should be of great interest.
Specifically, three research hypotheses must be evaluated.
Hypothesis 1. Reconfigurability can increase the perfor-
mance of supply chains, especially the resilience, flexibility,
and agility.
A quantitative proof of concept must be developed to eval-
uate the resilience, flexibility, and agility of reconfigurable
supply chains. As these performance indicators are difficult
to measure quantitatively and there are no benchmark
problems, the first step is to specify a framework for
benchmark problems concerning reconfiguration decisions.
Hypothesis 2. Knowledge graphs are an ideal basis for
data integration, decision support systems and AI deploy-
ment in supply chains.
Supply chains are characterized by multiple independent
units with incompatible and isolated information systems
and several data sources (Kache and Seuring, 2017; Zhang
et al., 2019). Knowledge graphs can potentially support
data integration and consequently the implementation
of decision support systems and AI in supply chains.
Compared to traditional relational databases, knowledge
graphs provide not only isolated and static data but also
context and logic. The presence of dynamic contextual
information enables reasoning because the arrangement
of data in knowledge graphs is reminiscent to the human
brain. It has the significant advantage that it is machine-
readable, human-understandable and allows structured
queries at the same time. The dissemination of knowledge
graphs for large-scale applications in big tech companies
also indicates that knowledge graphs could be promis-
ing tools for information management in supply chains,
especially because supply chains have a natural graph
structure.
Hypothesis 3. Graph-based AI has a higher solution qual-
ity and trustworthiness for strategic and tactical decision-
making in supply chain management than traditional
black-box AI.
As reconfiguration decisions are strategic decisions, the
risk of using traditional black-box AI is too high because
trustworthiness is more important for strategic and tacti-
cal decisions than decision-making time. Graph-based AI
algorithms have high synergies with knowledge graphs and
the natural graph structure of supply chains.
5. CONCEPTUAL AI SYSTEM
The concept considers supply chains as directed graphs
with nodes and arcs like in graph theory (See Figure
1). Each node represents a facility in the supply chain,
e. g. plants, warehouses, or customers, whereas each arc
represents a directed connection between two nodes, e. g.
material flow and/or information flow. Each reconfigura-
tion changes the graph structure by using one or more of
the following operations:
•Adding, removing, and replacing nodes
•Adding and removing arcs
•Reversing the arc direction
The AI system features a supply chain knowledge graph
that has two functions. On the one hand, it directly rep-
resents the structure of the supply network (blue nodes
and arcs in Figure 1). The weights on the arcs represent
the probabilities of establishing the respective supply re-
lation between two nodes. If the probability exceeds a
certain threshold (0.5 in Figure 1) the AI recommends
establishing the connection. On the other hand, the supply
chain knowledge graph includes additional information
(grey nodes and arcs in Figure 1) that is related to the
current state of the supply network, e. g. ongoing ship-
ments, disruptions, or master data. The grey nodes are the
basis for the graph-based AI to compute the probabilities
of the arcs. Basically, the knowledge graph is a digital
representation of the supply network that enriches it with
additional information. It is similar to the concept of the
digital twin which is a digital representation of a physical
object that exchanges data in both directions (Kritzinger
et al., 2018). The graph-based AI continuously gathers
information from the knowledge graph and dynamically
adjusts the weights, respectively reconfigures the system.
However, there are two differences to traditional digital
twins: Supply chains are no physical objects with physical
controls and reconfigurations require approval of a human
operator. Due to the importance of such decisions, the con-
cept works like a recommender system that also includes
the human in the loop. Whenever major changes occur in
the network, the graph AI recommends reconfigurations to
the human that he can choose from. The possible reconfig-
urations are ranked according to their weights and include
reasoning. The human can then choose his preferred option
and the graph-based AI receives feedback.
6. CONCLUSION
The COVID-19 pandemic has exposed the vulnerability
of modern supply chains and the need of fast and efficient
reactions in case of disruptions. Reconfigurability in supply
chains, which evolved from the concept of reconfigurable
manufacturing systems, pursues the objective of reducing
the time for strategic and tactical decisions, by introducing
dynamic reconfigurable services that allow a quick change
of supply relations. Although this concept is very promis-
ing, reconfigurable supply chains are still far from being
implemented in reality. Furthermore, literature is limited
to highly conceptual considerations that nevertheless re-
ceived a lot of attention. These concepts all agree that AI
algorithms are enablers for reconfigurable supply chains.
In this paper, we have highlighted the lack of precise
concepts for applying AI in reconfigurable supply chains,
and the need of benchmark problems of reconfigurable
supply chains on which the feasibility of AI methods can
be evaluated. To close both gaps, we have proposed a
human-centric AI approach based on a knowledge graph
that integrates the data of all participants in the supply
chain. As knowledge graphs can use the problem struc-
ture of reconfigurable supply chains, they are useful as
the backbone for the dynamic reconfiguration problem
on which more complex AI applications can be built.
The optimization algorithm gathers information from the
1.0
0.9
supplies
N1
0.1
N4 N5
0.6
manufactures
orders
name
N3
1.0
N2
requires
Bike
transport_mode material
Shipment
ScrewShip
XYZ
impacts
Disruption
0.2
0.9
supplies
N1
1.0
N4 N5
0.3
manufactures
orders
name
N3
1.0
N2
requires
Bike
transport_mode material
Shipment
ScrewShip
XYZ
impacts
Disruption
Feedback Loop
HumanGraph-based AI
...recommends reconfigurations based
on the supply chain knowledge graph
...takes the decision and gives feedback
to the graph-based AI
Supply Chain Knowledge Graph Reconfigured Supply Chain Knowledge Graph
Fig. 1. Reconfiguration process. A disruption impacts node 3 (N3) and triggers a potential reconfiguration. The graph-
based AI recalculates the probabilities for each supply relation. In this case, the algorithm recommends deactivating
the relations N1-N3 as well as N3-N5 and activate the relation N4-N5 instead. The human confirms the decision
and returns feedback to the AI algorithm.
knowledge graph and dynamically reconfigures the system
based on the inputs. It also utilizes the graph structure
of the reconfiguration problem by interpreting the nodes
and arcs as a trainable network whose weights can be
adjusted by backpropagation. Despite the commonly used
graph representation of supply chain problems, there are
only very few applications that use the inherent structure
of the supply chain or knowledge graphs. To enrich the
supply chain knowledge graph with contextual informa-
tion, the integration of open-source knowledge graphs such
as Wikidata or DBpedia is very promising. For example,
DBpedia, which is a crowd-sourced community project,
uses data from Wikipedia and currently comprises ap-
proximately 228 million entities (Lehmann et al., 2015).
Another interesting extension is the implementation of
real-time disruption monitoring with open databases. For
example, the GDELT project provides a global open-
source database of events gathered from broadcast, print
and web news (Leetaru and Schrodt, 2013). Online news-
papers report the causes for supply chain disruptions, for
instance natural disasters or strikes, usually very quickly.
Therefore, these data sources can significantly enhance the
data availability for the AI system and potentially improve
the reconfiguration decision-making. However, an integra-
tion of real-world data is only useful for real-world supply
chain problems but has limited usefulness for benchmark
problems.
REFERENCES
Ameri, F. and Kulvatunyou, B. (2019). Modeling a Supply
Chain Reference Ontology Based on a Top-Level On-
tology. In ASME 2019 International Design Engineer-
ing Technical Conferences and Computers and Informa-
tion in Engineering Conference. doi:10.1115/detc2019-
98278.
Aziz, A., Kosasih, E.E., Griffiths, R.R., and Brintrup, A.
(2021). Data Considerations in Graph Representation
Learning for Supply Chain Networks. arXiv.
Biswas, P., Kumar, S., Jain, V., and Chandra, C. (2019).
Measuring Supply Chain Reconfigurability using Inte-
grated and Deterministic Assessment Models. Jour-
nal of Manufacturing Systems, 52, 172–183. doi:
10.1016/j.jmsy.2019.05.008.
Bortolini, M., Galizia, F.G., and Mora, C. (2018). Recon-
figurable manufacturing systems: Literature review and
research trend. Journal of Manufacturing Systems, 49,
93–106. doi:10.1016/j.jmsy.2018.09.005.
Brintrup, A., Wichmann, P., Woodall, P., McFarlane, D.,
Nicks, E., and Krechel, W. (2018). Predicting Hidden
Benjamin Rolf et al. / IFAC PapersOnLine 55-10 (2022) 1693–1698 1697
1.0
0.9
supplies
N1
0.1
N4 N5
0.6
manufactures
orders
name
N3
1.0
N2
requires
Bike
transport_mode material
Shipment
ScrewShip
XYZ
impacts
Disruption
0.2
0.9
supplies
N1
1.0
N4 N5
0.3
manufactures
orders
name
N3
1.0
N2
requires
Bike
transport_mode material
Shipment
ScrewShip
XYZ
impacts
Disruption
Feedback Loop
HumanGraph-based AI
...recommends reconfigurations based
on the supply chain knowledge graph
...takes the decision and gives feedback
to the graph-based AI
Supply Chain Knowledge Graph Reconfigured Supply Chain Knowledge Graph
Fig. 1. Reconfiguration process. A disruption impacts node 3 (N3) and triggers a potential reconfiguration. The graph-
based AI recalculates the probabilities for each supply relation. In this case, the algorithm recommends deactivating
the relations N1-N3 as well as N3-N5 and activate the relation N4-N5 instead. The human confirms the decision
and returns feedback to the AI algorithm.
knowledge graph and dynamically reconfigures the system
based on the inputs. It also utilizes the graph structure
of the reconfiguration problem by interpreting the nodes
and arcs as a trainable network whose weights can be
adjusted by backpropagation. Despite the commonly used
graph representation of supply chain problems, there are
only very few applications that use the inherent structure
of the supply chain or knowledge graphs. To enrich the
supply chain knowledge graph with contextual informa-
tion, the integration of open-source knowledge graphs such
as Wikidata or DBpedia is very promising. For example,
DBpedia, which is a crowd-sourced community project,
uses data from Wikipedia and currently comprises ap-
proximately 228 million entities (Lehmann et al., 2015).
Another interesting extension is the implementation of
real-time disruption monitoring with open databases. For
example, the GDELT project provides a global open-
source database of events gathered from broadcast, print
and web news (Leetaru and Schrodt, 2013). Online news-
papers report the causes for supply chain disruptions, for
instance natural disasters or strikes, usually very quickly.
Therefore, these data sources can significantly enhance the
data availability for the AI system and potentially improve
the reconfiguration decision-making. However, an integra-
tion of real-world data is only useful for real-world supply
chain problems but has limited usefulness for benchmark
problems.
REFERENCES
Ameri, F. and Kulvatunyou, B. (2019). Modeling a Supply
Chain Reference Ontology Based on a Top-Level On-
tology. In ASME 2019 International Design Engineer-
ing Technical Conferences and Computers and Informa-
tion in Engineering Conference. doi:10.1115/detc2019-
98278.
Aziz, A., Kosasih, E.E., Griffiths, R.R., and Brintrup, A.
(2021). Data Considerations in Graph Representation
Learning for Supply Chain Networks. arXiv.
Biswas, P., Kumar, S., Jain, V., and Chandra, C. (2019).
Measuring Supply Chain Reconfigurability using Inte-
grated and Deterministic Assessment Models. Jour-
nal of Manufacturing Systems, 52, 172–183. doi:
10.1016/j.jmsy.2019.05.008.
Bortolini, M., Galizia, F.G., and Mora, C. (2018). Recon-
figurable manufacturing systems: Literature review and
research trend. Journal of Manufacturing Systems, 49,
93–106. doi:10.1016/j.jmsy.2018.09.005.
Brintrup, A., Wichmann, P., Woodall, P., McFarlane, D.,
Nicks, E., and Krechel, W. (2018). Predicting Hidden
1698 Benjamin Rolf et al. / IFAC PapersOnLine 55-10 (2022) 1693–1698
Links in Supply Networks. Complexity, 2018, 1–12. doi:
10.1155/2018/9104387.
Chandra, C. and Grabis, J. (2016). Supply Chain Con-
figuration, Concepts, Solutions, and Applications. doi:
10.1007/978-1-4939-3557-4.
Cˆandido, G., Barata, J., Colombo, A.W., and Jammes,
F. (2009). SOA in reconfigurable supply chains:
A research roadmap. Engineering Applications
of Artificial Intelligence, 22(6), 939–949. doi:
10.1016/j.engappai.2008.10.020.
Dolgui, A., Ivanov, D., and Sokolov, B. (2020). Recon-
figurable supply chain: the X-network. International
Journal of Production Research, 58(13), 1–26. doi:
10.1080/00207543.2020.1774679.
Ehrlinger, L. and W¨oß, W. (2016). Towards a Definition of
Knowledge Graphs. In Joint Proceedings of the Posters
and Demos Track of 12th International Conference on
Semantic Systems - SEMANTiCS2016 and 1st Inter-
national Workshop on Semantic Change & Evolving
Semantics (SuCCESS16), volume 1695.
Geerts, G.L. and O’Leary, D.E. (2014). A supply chain
of things: The EAGLET ontology for highly visible
supply chains. Decision Support Systems, 63, 3–22. doi:
10.1016/j.dss.2013.09.007.
Gomez-Perez, J.M., Pan, J.Z., Vetere, G., and Wu, H.
(2017). Enterprise Knowledge Graph: An Introduc-
tion. In Exploiting Linked Data and Knowledge Graphs
in Large Organisations, 1–14. doi:10.1007/978-3-319-
45654-6 1.
Grubic, T. and Fan, I.S. (2010). Supply chain ontology:
Review, analysis and synthesis. Computers in Industry,
61(8), 776–786. doi:10.1016/j.compind.2010.05.006.
Javorcik, B. (2020). Global supply chains will not be the
same in the post-COVID-19 world. In R.E. Baldwin
and S.J. Evenett (eds.), COVID-19 and Trade Pol-
icy: Why Turning Inward Won’t Work, 111–116. doi:
10.1111/twec.13238.
Kache, F. and Seuring, S. (2017). Challenges and op-
portunities of digital information at the intersection
of Big Data Analytics and supply chain management.
International Journal of Operations & Production Man-
agement, 37(1), 10–36. doi:10.1108/ijopm-02-2015-0078.
Kelepouris, T., Wong, C., Farid, A., Parlikad, A., and
McFarlane, D. (2006). Towards a Reconfigurable Sup-
ply Chain Model. In Intelligent Production Machines
and Systems, Intelligent Supply Chains, 481–486. doi:
10.1016/b978-008045157-2/50085-7.
Koren, Y., Heisel, U., Jovane, F., Moriwaki, T., Pritschow,
G., Ulsoy, G., and Brussel, H.V. (1999). Reconfigurable
Manufacturing Systems. CIRP Annals - Manufac-
turing Technology, 48(2), 527–540. doi:10.1016/s0007-
8506(07)63232-6.
Koren, Y., Gu, X., and Guo, W. (2018). Reconfigurable
manufacturing systems: Principles, design, and future
trends. Frontiers of Mechanical Engineering, 13(2), 121–
136. doi:10.1007/s11465-018-0483-0.
Kosasih, E.E. and Brintrup, A. (2021). A machine
learning approach for predicting hidden links in sup-
ply chain with graph neural networks. Interna-
tional Journal of Production Research, 1–14. doi:
10.1080/00207543.2021.1956697.
Kritzinger, W., Karner, M., Traar, G., Henjes, J., and
Sihn, W. (2018). Digital twin in manufacturing: A
categorical literature review and classification. IFAC
PapersOnLine, 51(11), 1016–1022.
Leetaru, K. and Schrodt, P.A. (2013). Gdelt: Global data
on events, location, and tone, 1979–2012. In ISA annual
convention, volume 2, 1–49. Citeseer.
Lehmann, J., Isele, R., Jakob, M., Jentzsch, A., Kon-
tokostas, D., Mendes, P.N., Hellmann, S., Morsey, M.,
Kleef, P.v., Auer, S., and Bizer, C. (2015). DBpedia
– A large-scale, multilingual knowledge base extracted
from Wikipedia. Semantic Web, 6(2), 167–195. doi:
10.3233/sw-140134.
L¨u, L. and Zhou, T. (2011). Link prediction in com-
plex networks: A survey. Physica A: Statistical Me-
chanics and its Applications, 390(6), 1150–1170. doi:
10.1016/j.physa.2010.11.027.
Qin, J., Liu, Y., and Grosvenor, R. (2016). A Cate-
gorical Framework of Manufacturing for Industry 4.0
and Beyond. Procedia CIRP, 52, 173–178. doi:
10.1016/j.procir.2016.08.005.
Ramzy, N., Auer, S., Chamanara, J., and Ehm, H.
(2021). Computer and Information Science 2021—Sum-
mer. Studies in Computational Intelligence, 61–75. doi:
10.1007/978-3-030-79474-3 5.
Schniederjans, D.G., Curado, C., and Khalajhedayati,
M. (2020). Supply chain digitisation trends: An in-
tegration of knowledge management. International
Journal of Production Economics, 220, 107439. doi:
10.1016/j.ijpe.2019.07.012.
Schroeder, M., See, B.v., Schnelle, J., and Kersten, W.
(2021). Impact of the Covid-19 Pandemic on Sup-
ply Chain Management. Logistik in Wissenschaft und
Praxis, 3–24. doi:10.1007/978-3-658-33480-2 1.
Singhal, A. (2012). Introducing the Knowledge Graph:
things, not strings.
Song, D., Schilder, F., Hertz, S., Saltini, G., Smiley, C., Ni-
varthi, P., Hazai, O., Landau, D., Zaharkin, M., Zielund,
T., Molina-Salgado, H., Brew, C., and Bennett, D.
(2017). Building and Querying an Enterprise Knowledge
Graph. IEEE Transactions on Services Computing,
12(3), 356–369. doi:10.1109/tsc.2017.2711600.
Zhang, W., Liu, Y., Jiang, L., Shah, N., Fei, X., and Cai,
H. (2019). The Construction of a Domain Knowledge
Graph and Its Application in Supply Chain Risk Anal-
ysis. Lecture Notes on Data Engineering and Commu-
nications Technologies, 464–478. doi:10.1007/978-3-030-
34986-8 33.
Zidi, S., Hamani, N., and Kermad, L. (2021a). Classi-
fication of Reconfigurability Characteristics of Supply
Chain. Lecture Notes in Mechanical Engineering, 72–
79. doi:10.1007/978-3-030-90700-6 7.
Zidi, S., Hamani, N., and Kermad, L. (2021b). Modularity
Metric in Reconfigurable Supply Chain. IFIP Advances
in Information and Communication Technology, 455–
464. doi:10.1007/978-3-030-85914-5 49.
Zidi, S., Hamani, N., and Kermad, L. (2021c). Reconfig-
urable Supply Chain Performance: A Bibliometric Anal-
ysis. IFIP Advances in Information and Communication
Technology, 161–169. doi:10.1007/978-3-030-85969-5 14.
Zou, X. (2020). A Survey on Application of Knowledge
Graph. Journal of Physics: Conference Series, 1487(1),
012016. doi:10.1088/1742-6596/1487/1/012016.
