Elsevier

Computers in Industry

Volume 125, February 2021, 103369
Computers in Industry

The main trends for multi-tier supply chain in Industry 4.0 based on Natural Language Processing

https://doi.org/10.1016/j.compind.2020.103369Get rights and content

Highlights

  • The coherence value comparison showed Mallet has better applicability than LDA in processing scientific articles’ abstracts.

  • The research about supply chains in Industry 4.0, indexed by SCI and SSCI from 2009 - 2018, are summarized.

  • Relationship diagrams disclosed the correlation among different parts and inspired readers to find suitable research ideas.

  • The analysis combining machine learning and review can be an excellent example of bridging subjective and objective analysis.

  • The main trends of multi-tier supply chains in Industry 4.0 were revealed.

Abstract

Multi-tier supply chains in Industry 4.0 are critical emerging issues today. This article briefly examines the Industry 4.0 policies in different countries. In order to decide on a better model and the number of topics in the model, a comparative test of the coherence value for two machine learning classification methods based on Latent Dirichlet Allocation was conducted. Subsequently, the article combines the traditional literature review method with a survey article referring to Industry 4.0 and multi-tier supply chain, indexed by science citation index expanded (SCI-EXPANDED) and social sciences citation index (SSCI) during 2009-2018. The research direction, research type, and research approaches of each paper were extracted, and the topics of all the articles were classified by machine learning, which provides feasible routes and valuable research directions for researchers in this field. Afterward, the research status and future research directions were identified. The combination of natural language processing in machine learning to classify research topics and traditional literature review to investigate article details greatly improved the objectivity and scientificity of the study and laid a solid foundation for further research.

Introduction

Industry 4.0 is a statement of the fourth industrial revolution, which refers to a new round of scientific and technological innovation. The concept of Industry 4.0 was first officially publicized by the German government in 2011(K. Zhou et al., 2015). Industry 4.0 is one of the more popular titles for the new round of industrial revolution. In addition to the German government, many other countries have also developed their own plans for the new series of scientific and technological reformations. France launched the project of “Usine du Futur” in 2013, then renamed the project as “Industrie du Futur” in 2015(FIM, 2015). The United States launched the Advanced Manufacturing Partnership (AMP) in 2011 (The White House, 2011). United Kingdom (UK) launched the Future of Manufacturing in 2013 (Government Office for Science, 2013). China put forward “Made in China 2025” in 2015 (L. Li, 2018). Not all countries define the new round of industrial revolution as the fourth industrial revolution. Japan established its plan as “The 5th Science and Technology Basic Plan” (Government of Japan, 2016); South Korea's program is called “Innovation in Manufacturing 3.0” (K. Lin et al., 2017). Table 1 shows the plans of different countries for a new round of industrial revolution. Since Industry 4.0 is currently the most commonly used term, in this article, we use Industry 4.0 to refer to a new industrial revolution.

As a focus point, Industry 4.0 has been studied and analyzed by many scholars. (Schneider, 2018) differentiated Industry 4.0 from digitalization and computer-integrated manufacturing to identified managerial challenges of Industry 4.0. (Alcácer and Cruz-Machado, 2019) evaluated new technologies that may appear in manufacturing systems, and projected to manufacturing systems composed by new technologies. (Kamble et al., 2018a) studied Industry 4.0 from a sustainable perspective and propose a sustainable Industry 4.0 framework based on the literature with Industry 4.0 technologies, process integration, and sustainable outcomes as the critical components of this framework. (Zhou and Cardinal, 2019) focused on the manufacturing sector and the financial and economic sectors under the background of Industry 4.0. They analyzed the development of these two areas and prospected to the future research direction, in which the supply chain was analyzed as an independent branch in the manufacturing sector, and considered that the multi-tier supply chain (MSC) would be one of the directions for future supply chain research.

The reviews for Industry 4.0 and supply chains are also emerging. (Ivanov et al., 2018b) constructed a survey for supply chain management and Industry 4.0, focusing on control theory applications to operational systems. (Saucedo-Martinez et al., 2018) examined categorically recent advances through qualitative and segmentation methods, making allowances for uncovering trends and areas of opportunity in the industry sense of 4.0 to attain research gaps that might be performed in organizational systems on the value chain. (Kamble et al., 2018b), based on the review of the literature from a sustainable perspective, proposed a sustainable Industry 4.0 framework consisting of three main components viz., Industry 4.0 technologies, process integration, and sustainable outcomes. (Dallasega, Rauch, et al., 2018) drew a literature review for construction supply chains using the concept of proximity to investigating synchronization between suppliers and the construction site and exhibit benefits and challenges for construction supply chains. (Ding, 2018) identified the potential sustainability barriers of the pharmaceutical supply chain and investigated how Industry 4.0 can be employed in the sustainable pharmaceutical supply chain paradigms.

In the review of the supply chain, a few scholars have reviewed the multi-tier supply chain. (Tachizawa and Wong, 2014) fused approaches and contingency variables to conduct the sustainability of multi-tier supply chains and sub-suppliers to establish a comprehensive framework. (Maestrini et al., 2017) contemplated a theory-testing approach regarding the relatively more mature component of supply chain performance measurement systems (SCPMSs) from a life cycle perspective along with a theory-building approach relating to the other under-investigated parts based on systematic literature review. (Tuni et al., 2018) explored in detail what the tiers of the supply chain are genuinely related to, in green performance assessment, which contributes to clarifying the range of the supply chain dimension in green supply chain management (GSCM) performance measurement research. (Zhu and Wang, 2018) identified the principal author collaborative networks and thematic trends of purchasing and supply management (PSM) by bibliometric review based on 371 peer-reviewed articles published between 1998 and 2017 using CiteSpace.

All of the above summarizes articles classified by Web of Science as review and indexed by SCI and SSCI related to “Industry 4.0 and Supply Chain” and “Multi-Tier Supply Chain” associated topics since 2009. The logic of the retrieval articles in this study will be detailed below. All the items are reviewed with the single exception of (Tachizawa and Wong, 2014) published after 2017. In other words, this is indeed a very new burning issue.

As stated in (Sarkis et al., 2019), sustainability issues pervade the supply chain deep into the recesses of various global regions and resources. Supply chains can evolve into quite complex systems as they form multiple tiers of organizations across networks. That is, the multi-tier supply chain referred to in this paper. Addressing these concerns is still in its relative infancy amongst business, engineering, and production economics solutions. In the meantime, as mentioned by (Lezoche et al., 2020), Industry 4.0 is about including and integrating the latest developments based on digital technologies as well as the interoperability process across them. However, the challenge remains are how to maintain these complex networked structures efficiently linked and organized within the use of such technologies, uniquely to identify and satisfy increasingly complex supply chain stakeholders’ dynamic requirements, are the challenges today. Meanwhile, (Wagire et al., 2020) pointed out as well, many manufacturing firms are looking forward to a significant impact of Industry 4.0 on their supply chains, operations, and business models. Nevertheless, complex characteristics of Industry 4.0 and the increasingly complex supply chain system are yet to be fully comprehended by most of them. Therefore, reviewing researches on the topic of combining Industry 4.0 and the multi-tier supply chain is important.

Besides, with the development of natural language processing (NLP), NLP has been widely practiced in semantic analysis, semantic classification, and sentiment analysis. By way of illustration, an adaptive two-stage feature selection approach for sentiment classification that applies to an arbitrary type of feature metrics and sentiment classifiers was presented by (Chi et al., 2017). (Cambria et al., 2018) lined sub-symbolic and symbolic AI to automatically detect conceptual primitives from text and link them to commonsense concepts and named entities in new three-level knowledge representation for sentiment analysis. (Kang et al., 2018) propose two novel NLP tasks based on a dataset of scientific peer reviews exhibited the simple models can predict whether a paper is accepted and predict the numerical scores of review aspects. (Wei Zhao et al., 2019) expanded existing capsule networks into a new framework with advantages concerning scalability, reliability, and generalizability. Their experimental results indicated its effectiveness on two NLP tasks of multi-label text classification and question answering. In terms of the review of text classification, (Minaee et al., 2020) reviewed more than 150 deep learning models evolved in the past 6–7 years and remarkably ameliorated state-of-the-art on various text classification tasks, containing sentiment analysis, news categorization, topic classification, question answering (QA), and natural language inference (NLI). This review article supplied an excellent peek into deep learning on text mining.

Nonetheless, the majority of current literature review studies mostly use a subjective evaluation-oriented systematic literature review method to carry out the classification of research topics. Subjective impression usually affects the objectivity of topic classification to some extent. As (Jurisica et al., 2001) said, although manual analysis has the advantage of flexibility, a computational approach makes the process scalable and more objective. Furthermore, (Sajda, 2006) also mentioned machine learning offers promise for improving the sensitivity and specificity of detection and diagnosis while at the same time increasing objectivity of the decision-making process, the applications of machine learning in their fields are very well-liked. (Jin et al., 2000) applied linear discriminant analysis based machine learning method to speaker recognition, the results showed that the accuracy of the model with linear discriminant analysis was, on average, 2.5 % higher than that without linear discriminant analysis, and the highest accuracy was 97 %. Same as other researchers, the objectivity referred to in this article is compared with the subjective judgment analysis that cannot be measured by quantitative objective criteria, and the higher accuracy is also in terms of the comparison experiments. Under such a comparison condition, we can say that machine learning-based methods are more objective than the traditional subjective classification.

On top of that, (Resch and Kaminski, 2019) pointed out that scientificity is essentially a methodology. Opacity and reproducibility are two significant keys to scientificity. Our study uses machine learning classification based on LDA and conducts coherence value comparison experiments for two models derived based on the Dirichlet distribution. According to the comparison results of objective coherence value, the better model, and the optimal number of topics are selected, which enhances the objectivity and transparency. Besides, the final result is obtained by using a fixed seed, which greatly enhances reproducibility. Therefore, compared with the classification based on subjective judgment, our methodology reduces opacity and significantly augments reproducibility. In short, the objectivity and scientificity of the study are improved. Hence, this paper is not only based on the traditional systematic literature review methods but also uses text mining to categorize the topics of articles by comparing different machine learning models to make the evaluation of literature more objective. In addition, the application of machine learning in the literature review will be discussed in detail in Section 2.2.

That is to say, at present, although some scholars have reviewed topics related to the sustainable/global supply chain, few scholars have reviewed the literature of the multi-tier supply chain combining with machine learning. What is more, although some scholars have investigated “Industry 4.0 and supply chain”, few scholars perform an investigation combining Industry 4.0 and the multi-tier supply chain. This paper will focus on multi-tier supply chain (MSC) research and combine Industry 4.0 to explore the research status and future development direction of MSC in the context of Industry 4.0, so as to fill the gaps in the current research. Additionally, most of the current review articles are based on manual and subjective analysis methods of systematic literature review methods such as bibliometrics. The combination of traditional research methods and the use of more objective and advanced research methods will also greatly enhance the innovativeness and practicality in this new research field.

Section snippets

Methodology

The primary purpose of this study is to fill the gaps in the research of Industry 4.0 and multi-tier supply chain, analyze the research status, and look forward to future research directions. As shown in Fig. 1, this paper will be divided into four parts.

Text mining

One of the main applications of natural language processing is to automatically extract topics that people are discussing from large amounts of text. As (Wang et al., 2015) say, LDA was developed to overcome the problems of data sparsity and ambiguity that the traditional text mining method is based on bag-of-words (BoW) in short text modeling. Therefore, this paper uses LDA based methods to classify the retrieved articles using the abstract of the 157 articles retrieved according to the above

Manual analysis

Since the accuracy of the existing natural language processing technology cannot reach the level of manual analysis, manual analysis is used in this part to make up for the shortage of text mining.

Conjoint analysis

The text mining part above has given the topic classification based on LDA based Mallet model, and the manual analysis part has analyzed the current research paths in Industry 4.0 and multi-layer supply chain in detail. The joint analysis above will be carried out in this part. We set the topic with the highest percentage contribution in the text mining section as the topic of this article.

Future opportunities and challenges

According to the above analysis, there are few high-quality quantitative studies aimed at corporate strategies to adapt to the background of Industry 4.0, which is still in the initial stage where data collection has just begun. As more and more data is collected, using quantitative research will significantly improve the rigor of the research. What is more, when the amount of data reaches a certain level, methods such as deep learning combined with big data will profoundly affect the reform of

Conclusion

This article first summarized the current new round of industrial revolution plans in different countries, then described the status of literature review on multi-tier supply chain and Industry 4.0, afterward reviewed papers on multi-tier supply chain and Industry 4.0 in the past ten years indexed by SSCI and SCI, after that two types of text mining methods, LDA and Mallet, were used to classify the topics. Mallet models with higher coherence values were selected to sort the literature into

Declaration of Competing Interest

The authors report no declarations of interest.

Acknowledgments

Thanks, human resources, material, and financial support provided by the Université Paris-Saclay, CentraleSupélec, and Concordia University. Furthermore, we would like to extend our sincere gratitude to the funds provided by the Canada Mitacs (FR37639) and the China Scholarship Council (JXJZHM 20171324) to support our research.

Rongyan Zhou obtained his Bachelor's degrees in Industrial Engineering from Beihang University, China, in 2016, and received his Master's degree in Complex Systems Engineering from Université Paris-Saclay, CentraleSupélec, France, in 2018. His current research interests concentrate on the impact of industry 4.0 on economics and management, particularly focusing on the application of artificial intelligence in economics and management. He is currently a Ph.D. student within the research group of

References (201)

  • C.-F. Chien et al.

    Strategic capacity planning for smart production: decision modeling under demand uncertainty

    Appl. Soft Comput.

    (2018)
  • M. Conde-Cid et al.

    Degradation of sulfadiazine, sulfachloropyridazine and sulfamethazine in aqueous media

    J. Environ. Manage.

    (2018)
  • P. Dallasega et al.

    Understanding greening supply chains: proximity analysis can help

    Resour. Conserv. Recycl.

    (2018)
  • P. Dallasega et al.

    Industry 4.0 as an enabler of proximity for construction supply chains: a systematic literature review

    Comput. Ind.

    (2018)
  • D. Delen et al.

    Seeding the survey and analysis of research literature with text mining

    Expert Syst. Appl.

    (2008)
  • K. Demeter et al.

    A text mining based overview of inventory research in the ISIR special issues 1994–2016

    Int. J. Prod. Econ.

    (2019)
  • G.D. DeYong et al.

    Is dishonesty the best policy? Supplier behaviour in a multi-tier supply chain

    Int. J. Prod. Econ.

    (2015)
  • B. Ding

    Pharma Industry 4.0: literature review and research opportunities in sustainable pharmaceutical supply chains

    Process. Saf. Environ. Prot.

    (2018)
  • Y. Dou et al.

    Green multi-tier supply chain management: an enabler investigation

    J. Purch. Supply Manag.

    (2018)
  • F. Dunke et al.

    Time traps in supply chains: Is optimal still good enough?

    Eur. J. Oper. Res.

    (2018)
  • M.A. Edalatpour et al.

    Investigation on a novel sustainable model for waste management in megacities: a case study in tehran municipality

    Sustain. Cities Soc.

    (2018)
  • F. Elgh

    Decision support in the quotation process of engineered-to-order products

    Adv. Eng. Inform.

    (2012)
  • F. Galati et al.

    Industry 4.0: emerging themes and future research avenues using a text mining approach

    Comput. Ind.

    (2019)
  • Joerg H. Grimm et al.

    Critical factors for sub-supplier management: a sustainable food supply chains perspective

    Int. J. Prod. Econ.

    (2014)
  • J.örg H. Grimm et al.

    Exploring sub-suppliers’ compliance with corporate sustainability standards

    J. Clean. Prod.

    (2016)
  • Jorg H. Grimm et al.

    Interrelationships amongst factors for sub-supplier corporate sustainability standards compliance: an exploratory field study

    J. Clean. Prod.

    (2018)
  • V. Gruzauskas et al.

    Minimizing the trade-off between sustainability and cost effective performance by using autonomous vehicles

    J. Clean. Prod.

    (2018)
  • L.A. Haidari et al.

    The economic and operational value of using drones to transport vaccines

    Vaccine

    (2016)
  • K. Holvoet et al.

    Quantitative study of cross-contamination with Escherichia coli, E. coli O157, MS2 phage and murine norovirus in a simulated fresh-cut lettuce wash process

    Food Control

    (2014)
  • X. Huang et al.

    Consideration and research of high voltage insulation strategy for ITER Feeder busbar joint

    Fusion Eng. Des.

    (2013)
  • J. Hultman et al.

    An interaction approach to global sourcing: a case study of IKEA

    J. Purch. Supply Manag.

    (2012)
  • A.J. Isaksson et al.

    The impact of digitalization on the future of control and operations

    Comput. Chem. Eng.

    (2018)
  • D. Ivanov et al.

    A survey on control theory applications to operational systems, supply chain management, and Industry 4.0

    Annu. Rev. Control

    (2018)
  • S.S. Kamble et al.

    Sustainable Industry 4.0 framework: a systematic literature review identifying the current trends and future perspectives

    Process. Saf. Environ. Prot.

    (2018)
  • S.S. Kamble et al.

    Analysis of the driving and dependence power of barriers to adopt industry 4.0 in Indian manufacturing industry

    Comput. Ind.

    (2018)
  • J. Kembro et al.

    Information sharing across multiple supply chain tiers: a Delphi study on antecedents

    Int. J. Prod. Econ.

    (2017)
  • F.C. Kleemann et al.

    A providers’ perspective on supplier relationships in performance-based contracting

    J. Purchasing Supply Manage.

    (2013)
  • L. Knight et al.

    Supply market uncertainty: exploring consequences and responses within sustainability transitions

    J. Purchasing Supply Manage.

    (2015)
  • R. Kuenzel et al.

    SmartSite: intelligent and autonomous environments, machinery, and processes to realize smart road construction projects

    Autom. Construct.

    (2016)
  • I. Akef et al.

    Mallet vs GenSim: topic modeling for 20 news groups report

    Univ. Ark. Little Rock Law J.

    (2016)
  • Anonymous

    Rusatom overseas names key sub-suppliers for fennovoimas’ hanhikivi NPP

    Atw – Int. J. Nuclear Power

    (2015)
  • C.B. Asmussen et al.

    Smart literature review: a practical topic modelling approach to exploratory literature review

    J. Big Data

    (2019)
  • L. Avelar-Sosa et al.

    Structural model to assess the relationship of manufacturing practices

    South Afr. J. Ind. Eng.

    (2018)
  • T. Banyai

    Real-time decision making in first mile and last mile logistics: how smart scheduling affects energy efficiency of hyperconnected supply chain solutions

    Energies

    (2018)
  • T. Banyai et al.

    Smart scheduling: an integrated first mile and last mile supply approach

    Complexity

    (2018)
  • J. Barata et al.

    Mobile supply chain management in the Industry 4.0 era an annotated bibliography and guide for future research

    J. Enterp. Inf. Manag.

    (2018)
  • L. Bibby et al.

    Defining and assessing industry 4.0 maturity levels - case of the defence sector

    Prod. Plan. Control.

    (2018)
  • F. Bienhaus et al.

    Procurement 4.0: factors influencing the digitisation of procurement and supply chains

    Bus. Process. Manag. J.

    (2018)
  • S. Bistarelli et al.

    On merging two trust-networks in one with bipolar preferences

    Math. Struct. Comput. Sci.

    (2017)
  • D.M. Blei et al.

    Latent Dirichlet allocation

    J. Mach. Learn. Res.

    (2003)
  • Cited by (0)

    Rongyan Zhou obtained his Bachelor's degrees in Industrial Engineering from Beihang University, China, in 2016, and received his Master's degree in Complex Systems Engineering from Université Paris-Saclay, CentraleSupélec, France, in 2018. His current research interests concentrate on the impact of industry 4.0 on economics and management, particularly focusing on the application of artificial intelligence in economics and management. He is currently a Ph.D. student within the research group of Advanced Manufacturing and Industry 4.0 in Industrial Engineering Laboratory (LGI) at CentraleSupélec, Université Paris-Saclay, France, since November 2017.

    Dr. Anjali Awasthi is Associate Professor and Concordia University Research Chair (Tier-II) in Connected Sustainable Mobility Systems at Concordia Institute for Information Systems Engineering (CIISE), in Concordia University, Montreal. She received a PhD in industrial engineering and automation from INRIA Rocquencourt and University of Metz, France and a Masters in Industrial and Management Engineering from IIT Kanpur, India. Prior to Concordia, Dr. Awasthi worked at University of British Columbia and University of Laval where she was involved in several projects on industrial applications of operations research. In France, she was involved in many European projects aimed at improving urban mobility in cities, city logistics and on cybernetic transportation systems. Her areas of research are modeling and simulation, data mining, Information Technology and decision making, sustainable logistics planning, quality assurance in supply chain management and sustainable supply chain management. She is currently serving as the Education Chair for CORS (Canadian Operations Research Society), a senior member of ASQ (American Society for Quality), associate of LSRC (Loyola Sustainability Research Center), and regular member of CIRRELT (Centre Interuniversitaire de Recherche sur les Reseaux d'Entreprise, la Logistique et le Transport). She is also the recipient of Eldon Gunn service award (CORS 2018, Halifax) and IEOM Special Recognition Award (4th North American Conference on Industrial Engineering and Operations Management, Toronto, 2019).

    Julie Stal-Le Cardinal is professor in the Industrial Engineering Laboratory (LGI) at CentraleSupélec, France. Her research interest is to discover new methodologies of design and management in an industrial context. She works on team diversity and project management. She coaches project managers in companies concerning the management of their project and the choice of actors in teams. Within the LGI, she is in charge of two research groups: one about Industrial Engineering & Healthcare (GIM’S) and one about Advanced Manufacturing and Industry 4.0. After a Mechanical Engineer Degree (UTC, France), she had a Master Degree in Industrial Engineering Systems, at ECP, and her PhD Thesis delivers A Study of Dysfunctions within the Decision Making Process. Particular Focus on the Choice of Actor. Prof. Julie Stal-Le Cardinal is now Head of Enterprise Science Department at CentraleSupélec. Mechanical engineer (University of Technology of Compiègne, France): Industrial Design, Project methodologies (Value Analysis and Project Management (applications on concrete and varied industrial cases)). Master of Industrial Engineering Systems, at the Ecole Centrale Paris, with a report on capitalization of the know-how in conception. PhD Thesis: A Study of Dysfunctions within the Decision Making Process. Particular Focus on the Choice of Actor. The thesis concerns the optimisation of the decision process in industrial projects: observation of various industrial projects. Validation of the proposed models within the framework of a training at Vallourec (meetings with project managers, to analyse dysfunctions in their projects and proposition of action plan and recommendations). Habilitation Thesis: Systemic Approach for Decision Making in Industry. (https://hal.archives-ouvertes.fr/tel-00786203v1). web: https://www.centralesupelec.fr/fr/laboratoire-genie-industriel-lgi-ea-2606. Date of Birth: 29/05/1973.

    View full text