Bibliometric and Text Analytics Approaches to Review COVID-19 Impacts on Supply Chains

Abstract

The current COVID-19 pandemic has virtually disrupted supply chains worldwide. Thus, supply chain research has received significant attention. While the impacts have been immeasurable, organizations have realized the need to design strategies to overcome such unexpected events. Therefore, the supply chain research landscape has evolved to address the challenges during the pandemic. However, available literature surveys have not explored the power of text analytics. Hence, in this review, an analysis of the supply chain literature related to the impacts of COVID-19 is performed to identify the current research trends and future research avenues. To discover the frequent topics discussed in the literature, bibliometric analysis (i.e., keyword co-occurrence network) and text mining tools (i.e., N-gram analysis and topic modeling) are employed for the whole corpus and the top-three contributing journals (i.e., Sustainability, International Journal of Logistics Management, Operations Management Research). Moreover, text analytics (i.e., Term Frequency-Inverse Document Frequency: TF-IDF) is utilized to discover the distinctive topics in the corpus and per journals. A total of 574 papers published up to the first semester of 2022 were collected from the Scopus database to determine the research trends and opportunities. The keyword network identified four clusters considering the implementation of digitalization to achieve resilience and sustainability, the usage of additive manufacturing during the pandemic, the study of food supply chains, and the development of supply chain decision models to tackle the pandemic. Moreover, the segmented keyword network analysis and topic modeling were performed for the top three contributors. Although both analyses draw the research concentrations per journal, the keyword network tends to provide a more general scope, while the topic modeling gives more specific topics. Furthermore, TF-IDF scores unveiled topics rarely studied, such as the implications of the pandemic on plasma supply chains, cattle supply chains, and reshoring decisions, to mention a few. Additionally, it was observed how the different methodologies implemented allowed to complement the information provided by each method. Based on the findings, future research avenues are discussed. Therefore, this research will help supply chain practitioners and researchers to identify supply chain advancements, gaps in the literature and future research streams.

1. Introduction

The COVID-19 pandemic has impacted all sectors of the economy across the world, demonstrating the vulnerable supply chain environment [1]. The measures taken to mitigate the transmission of the virus, such as wearing protective equipment, quarantining, lockdowns, and many more, have negatively impacted operations and supply chains across all domains. Supply chain (SC) disruptions have been caused by both supply and demand [2,3]. Traditional SC strategies that have helped manufacturers perform efficiently were questioned during the pandemic. For instance, the just-in-time policy, which aimed at having a low stock of finished goods and raw material inventory, helped cut costs. However, when the demand for items such as toilet paper skyrocketed, the firms faced production problems due to a shortage of raw material inventory. Multinational firms experienced a supply shock; for example, face mask exports halted as the virus spread throughout India. There was a surge in demand for necessities, but there were also worries about delivery delays, delays in receiving goods, unanticipated travel delays, and labor scarcity. As a result, the disparity between supply and demand widened [2].

The pandemic impacts have not only been observed on multinational companies but also on Small & Medium Sized Enterprises (SMEs) and Micro, Small & Medium Enterprises (MSMEs) globally. For instance, Melnyk, et al. [4] studied SME suppliers to the U.S. Department of Defense; they highlighted that suppliers should not be treated as a homogenous subset and recommended expanding the supplier base. Similarly, Le, et al. [5] studied the effects on SMEs in a province in Vietnam; they also proposed strategies that the government could use to help SMEs to recover from the pandemic. Although different governments tried to implement measures to counter the impacts, the strategies must be tailored to the organization’s scale. Moreover, countries and regions where the SC entities are located need to be considered, as the industries’ characteristics and government policies could vary and result in different consequences [6].

Over the past few years, a lot of research has been done on the effects of the pandemic on SCs globally. Researchers have also suggested various operations research and SC strategies to combat the ill effects of the pandemic. Sarkis, et al. [7] indicated that SC managers should identify local sources that provide the same quality and quantity of raw materials and reshape their global sourcing. For instance, Corominas [8] developed a procurement-inventory system with the optimal combination of permanent stock, production capacity, domestic production, and imports to provide essential goods to the public. In the same vein, to increase the resilience of SCs delivering personal protective equipment (PPE), Ash, et al. [9] provided a multi-period, multi-objective, distributionally robust optimization framework.

While there has been a lot of attention studying SC during the pandemic, future researchers must understand the different discussion topics and future research avenues, which can be done by reviewing the literature. There have been studies that provide a systematic review related to the effects of the pandemic on SCs, but each has limitations that need to be addressed. Moosavi, Fathollahi-Fard and Dulebenets [3] have focused mainly on conducting a bibliometric analysis to find the most influential authors using citations. Since the recent papers would have less number of citations than the papers published earlier, the results would keep varying over time. Swanson and Santamaria [10] did a similar study but submitted their publication in December 2020, covering about 200 journal articles. However, there has been a significant increase in this domain’s publications since then. Sombultawee, et al. [11] have studied an updated database but limited the study to supply chain management (SCM) and VOSviewer features. Recently, Ardolino, et al. [12] presented a systematic literature review; however, it was from the operations management perspective.

Literature reviews analyzing COVID-19 literature based on the text analytics approach have been scarcely presented. For instance, Anderson [13] developed a literature survey using a text-mining clustering approach to study only abstracts from research articles related to COVID-19. However, that study did not emphasize the impacts on SCs. Meyer, et al. [14] studied the effect of the pandemic on SCs and their sustainability using text mining (e.g., word clouds, frequency plots, and sentiment analysis) by analyzing general newspapers and SC and logistics newspapers. Nevertheless, they did not study scientific journals. Similarly, Carracedo, et al. [15] studied the research lines of the pandemic’s impacts for businesses and marketing areas using text analytics.

State-of-the-art reviews based on bibliometric and data analytics have been explored in other domains. For instance, Morashti, et al. [16] studied the use of sustainable packaging in SC management. Similarly, Andersen, et al. [17] employed bibliometric and sentiment analysis to discover the most frequent topics and highlight trends in research for different scientific conferences which demonstrated the venues in which researchers published their findings according to the suitability of the conference.

From the extant literature, it is evident that systematic reviews based on qualitative data have been widely presented. However, surveys using quantitative techniques such as text mining and analytics have been scarcely proposed. Moreover, to the best of our knowledge, the simultaneous analysis of bibliometric and text analytics has not been used for studying the pandemic’s impacts on SCs. The capabilities of these methodologies can provide more insights than traditional approaches, which depend on the researcher’s bias and qualitative information. The previously mentioned particulars have been identified as the research gap and a domain to further research. Therefore, in this paper, bibliometric and text analytics methods are adopted to understand the literature highlighting the effects of the pandemic on SCs and different strategies used to combat this effect.

Bibliometric analysis was performed; specifically, a keyword co-occurrence network was created, and cluster analysis was carried out. Moreover, text analytics was conducted by implementing title analysis and topic modeling. Furthermore, Term Frequency-Inverse Document Frequency (TF-IDF) scores were calculated to unveil unique topics. Bibliometric and text analytics capabilities could complement each other and unveil meaningful patterns and new insights that other methodologies cannot discover. Therefore, our paper aims to answer the following research questions:

RQ1:

What are the different SC research avenues observed during the COVID-19 pandemic?

RQ2:

What research trends are studied by the top contributing journals?

RQ3:

What are the SC tools used to recover from the pandemic impacts?

RQ4:

What are the opportunities for future research?

The organization of this paper is as follows. Section 2 describes the research design of this study, while Section 3 describes the bibliometric analysis. Section 4 reports the results of the text analytics study. Section 5 discusses the theoretical and managerial implications. Section 6 contains the future research avenues, and Section 7 concludes with a summary of findings and states the limitation of this study.

2. Research Design

This review is conducted using the methodology framework shown in Figure 1. An initial exploration of the literature was performed to establish the research background and define the search terms for collecting the data, as mentioned in Section 2.1. Then, the data was prepared (e.g., lemmatization and stemming). Afterward, the analyses were executed for the entire corpus and specific journals. The keyword co-occurrence networks were created to provide a perspective on the frequent research topics discussed in the literature and in the top three contributing journals. From the text mining approach, title analysis was carried out for the complete corpus and topic modeling for the selected journals. Finally, text analytics was implemented via TF-IDF scores to identify the distinctive research topics per journal. Grounded on the findings, a discussion of future research is presented.

2.1. Defining Keywords for Search

This research is aimed at exploring the different perspectives in which the current COVID-19 pandemic has impacted the SC domain. Therefore, a search in Scopus was carried out on 13 June 2022. Articles containing the keywords “supply chain” AND “COVID” in the title, abstract, or keywords were selected. Although there is no specific setting about the year of publications, filters such as English articles and peer-reviewed documents were applied to obtain the resulting 574 articles. The Scopus database was used in this study as it is known to be 60% larger than the Web of Science database [18].

2.2. Search Results

The database consists of 574 published articles from 34 different journals, the count of papers from each journal is depicted in Figure 2. The highest number of papers came from the Sustainability journal, with an impact factor of 3.889 and a total count of 103 papers. The next highest contributing journal is the International Journal of Logistics Management (IJLM), with a contribution of 42 articles and an impact factor of 5.446, followed by Operations Management Research (OMR), with a total count of 27 papers and an impact factor of 7.032. From the number of articles published per journal, it can be noted that most authors preferred Sustainability, as it is open access and takes approximately 20 days for the first decision of peer review, which could be convenient to accelerate the research dissemination due to the nature of the topic.

As mentioned in the methodology framework, the analyses were performed on the entire corpus. Furthermore, the top three contributing journals (i.e., Sustainability, IJLM, and OMR) were analyzed.

2.3. Bibliometric Analysis

Bibliometric analysis is a statistical method to identify emerging trends by gathering and analyzing a large volume of scientific data [19]. Hence, bibliometric analysis is used to identify the major topics relating to SC research during the COVID-19 pandemic. One of the most common bibliometric methods is the analysis of the frequency of keywords co-occurrence [20,21]. Due to the fact that keywords reflect the main content of the articles, the keyword network can unveil the main topics accurately and minimize the bias in judgment [22,23].

Co-occurrence analysis is the relation between different words in the corpus based on the number of times they are mentioned in the document. The keyword co-occurrence network was created using VOSviewer (www.vosviewer.com, accessed on 21 July 2022), a software used to construct and view various bibliographic maps [24]. The keyword network contains circles and links. The circle size represents the weight of each keyword (i.e., frequency), and the proximity of the keywords in the network shows their relatedness. Moreover, the links’ strength is represented by the thickness of the lines.

The author keywords and index keywords were used to create the co-occurrence network. The keywords that were not in the root form were subject to lemmatization to avoid repetition in the network. A full counting was considered to ensure weights were similar for each co-occurrence. Only words occurring at least ten times in the document were included in the analysis.

2.4. Text Mining and Analytics

Text mining is a method that is a precursor to text analytics [25]. Text mining uses natural language processing, data mining, knowledge management, and machine learning techniques to process text documents and identify interesting and non-trivial information from unstructured text [26]. Similarly, text analytics combines machine learning, statistical, and linguistic techniques to extract and generate useful non-trivial information and knowledge [27]. Text mining and analytics tools have many applications and are used across various industries, like media, telecommunications, and healthcare.

2.4.1. N-Gram Analysis for Titles

In the fields of computational linguistics and probability, an N-gram is a continuous sequence of $N$ elements from a given sample of text or speech. The N-gram technique was applied to the paper titles in the database to understand the most frequently occurring contiguous words. The formula obtained by Jurafsky and Martin [28], as shown in Equation (1), was used to find the most frequent bigrams.

$$\mathrm{P}(x_N|x_{N-1})=\frac{\mathrm{r}\left(x_{N-1}{x}_N\right)}{{{\displaystyle \sum }}_{x}r\left(x_{N-1}{x}_N\right)}$$

(1)

where $x_{N-1}$ and $x_N$ are the words in review, $\mathrm{P}(x_N|x_{N-1})$ is the probability of $N$ words occurring in a sequence, and $\mathrm{r}\left(x_{N-1}{x}_N\right)$ is the ratio count of $x_{N-1}$ and $x_N$.

The paper titles were subject to a set of pre-text processing steps, as shown in Figure 3—lemmatization of the words ensured that different forms of the same word were not repeated. The keywords used in our initial search were removed to avoid any bias in our results.

2.4.2. Topic Modeling

Topic modeling is a technique in natural language processing used to identify the different topics in a corpus. It is an unsupervised technique with a wide array of applications, such as marketing, sociology studies, discourse analysis, and many more, used to identify underlying topics [29].

In this paper, a topic modeling tool called Latent Dirichlet Allocation (LDA) from the SciKit library [30] was used to analyze the different topics observed in the paper abstracts. The LDA model is a Bayesian probabilistic model which groups the words into the required number of topics. Hoffman, et al. [31] proposed an online variational Bayes algorithm for LDA, which is the base behind the topic modeling used in this paper. The abstracts were subject to text pre-processing, as given in Figure 3. The cleaned text was passed through the model to extract four topics from each of the top 3 contributing journals, the extracted topics are provided in Section 4.2.

2.4.3. Term Frequency—Inverse Document Frequency

This paper employs TF-IDF, a statistical tool that analyzes each word’s importance in a corpus. Term Frequency (TF) describes how often a specific word is repeated in a sentence and Inverse Document Frequency (IDF) is the inverse of the number of documents divided by the number of documents the word occurs in. It gives the importance of a particular word in the given corpus. Ramos [32] has provided a mathematical framework to calculate the TF-IDF score. The formula can be simplified into two halves, TF and IDF, as shown in Equations (2) and (3).

$$TF=\frac{No. of times the word occurs in a sentence}{No. of words in the sentence}$$

(2)

$$IDF=\log (\frac{No. of sentences in the document}{No. of sentences containing the word})$$

(3)

Finally, the TF-IDF score is calculated by multiplying Equations (2) and (3).

$$TF-IDF Score=\frac{No. of times the word occurs in a sentence}{No. of words in the sentence}\times \log (\frac{No. of sentences in the document}{No. of sentences containing the word})$$

(4)

The higher the score, the more distinctive a word would be to a particular document. Before calculating the TF-IDF scores, the abstracts were subject to text pre-processing tools depicted in Figure 3. The processed text went through the TF-IDF vectorizer from the SciKit learn library [30] in Python programming language. Then, the dataset was grouped according to journals to analyze each journal’s publication in-depth. The TF-IDF scores were used to identify the most unique topics discussed in each journal. Afterwards, TF-IDF scores for the complete corpus were calculated to provide general insights for the research community regarding possible research avenues.

3. Bibliometric Analysis

The impact of the publications can be assessed by the number of citations that an article receives. However, it is relevant to highlight that older articles could accumulate more citations than newer articles, and the number of citations will fluctuate over time. On

Table 1. Top ten most cited articles.

Title	Summary	Ref.	Citations
Predicting the impacts of epidemic outbreaks on global supply chains: A simulation-based analysis on the coronavirus outbreak (COVID-19/SARS-CoV-2) case	The paper focuses on the impacts of an epidemic on SCs. They propose a simulation-based methodology to identify and predict the effects of epidemics on SCs. They aim to address three features of epidemic outbreaks, namely: (1) Long-term disruption existence and its unpredictable scaling, (2) Simultaneous disruption propagation (i.e., the ripple effect) and epidemic outbreak propagation (i.e., pandemic effect), and (3) simultaneous disturbances in supply, logistics infrastructure and demand.	Ivanov [33]	673
Viability of intertwined supply networks: extending the supply chain resilience angles towards survivability. A position paper motivated by COVID-19 outbreak	The paper focuses on conceptualizing a decision-making environment for SC resilience by combining both intertwined supply networks (ISN) and viability. With COVID-19 outbreak also brings a new perspective to SC resilience research by foreseeing disruptions. They also designed a viability concept idea that resembles the ISN through dynamic game-theoretic modeling of a biological system.	Ivanov and Dolgui [34]	457
Food supply chains during the COVID-19 pandemic	The paper analyzes the COVID-19 impacts on the food SC from supply and demand perspectives. Some of the demand side issues were due to panic buying and changes in consumption patterns. The supply side issues involved labour shortages due to illness or self-monitoring, disruption in transportation network and the border restriction in moving goods. Lastly, it talks about the viability of online grocery retail and local food SC in the post-pandemic scenario.	Hobbs [35]	425
Viable supply chain model: integrating agility, resilience and sustainability perspectives—lessons from and thinking beyond the COVID-19 pandemic	A SC is considered viable based on three perspectives: agility, resilience, and sustainability. To create SCs that respond adaptively to positive changes, absorb negative disturbances, recover, and survive during long and short-term disruptions and global shocks with societal and economic transitions, decision-makers may find value in the viable SC model. Finally, future research directions in SC viability are specified.	Ivanov [36]	296
Impacts of epidemic outbreaks on supply chains: mapping a research agenda amid the COVID-19 pandemic through a structured literature review	This paper follows a systematic literature review to understand the epidemic’s impacts on commercial SCs. It was found that the most popular topic was related to optimizing resource allocation and distribution. Gaps in literature were grouped into three clusters: modeling, technology and organizational. A framework was proposed to tackle the COVID-19 pandemic regarding adaptation, digitalization, preparedness, recovery, ripple effect, and sustainability.	Queiroz, et al. [37]	258

and

Table 2. Top ten most cited articles (continuation).

Title	Summary	Ref.	Citations
A decision support system for demand management in healthcare supply chains considering the epidemic outbreaks: A case study of coronavirus disease 2019 (COVID-19)	This paper proposes a decision support system for demand management in healthcare SCs considering the impacts of the COVID-19 pandemic. Many communities face a shortage of medical and human resources under such conditions due to the high rate of disease outbreaks. The decision system is based on the physician’s knowledge and the Fuzzy Inference System. Finally, the model was tested on a real-world problem and was very efficient and accurate.	Govindan, et al. [38]	213
Research opportunities for a more resilient post-COVID-19 supply chain—closing the gap between research findings and industry practice	This paper aims to improve SC resilience literature by conducting interviews and surveys with experts in the industry. It helps identify the challenges existing in the industry and areas where research can support efforts in the industry. The challenges were grouped into supply, demand, and control risks and finally ended with the measures taken by participating executives to improve resilience.	Remko [39]	192
Impact of COVID-19 on logistics systems and disruptions in the food supply chain	This study aims to develop a simulation model of a public distribution system by varying demand across three scenarios that demonstrate disruptions in the food SC due to COVID-19. It focuses mainly on the mitigation strategies that can be implemented to manage disruptions in the logistics system of the food and healthcare SC.	Singh, et al. [40]	186
COVID-19 pandemic related supply chain studies: A systematic review	This paper analyzes SCs’ impacts during the initial stages of the pandemic. It selected 74 research articles and identified four research themes. It also mentions that most research was related to healthcare and high-demand goods, while low-demand goods and SMEs have been ignored.	Chowdhury, et al. [41]	152
Global socio-economic losses and environmental gains from the coronavirus pandemic	This paper aims to analyze the world’s socioeconomic losses and environmental gains from the pandemic. One of the environmental gains was the reduction in air pollution, as the consumption of fossil fuels was reduced due to travel restrictions. Socially, it resulted in labor shocks as migrant and unskilled laborers could not perform their work virtually.	Lenzen, et al. [42]	137

, the summary of the top ten cited articles is presented. The most cited article at the moment of collecting the data is entitled “Predicting the impacts of epidemic outbreaks on global supply chains: A simulation-based analysis on the coronavirus outbreak (COVID-19/SARS-CoV-2) case,” which presents a simulation model to evaluate the impacts of epidemics at a global scale. They highlighted the timing of closing and opening SC facilities would have a greater effect on the SC performance than the disruption in upstream echelons or its propagation.

The primary objective of the ten most cited articles is to understand the impacts of the pandemic on global SCs, food SCs, and healthcare SCs. Literature reviews, simulations, conceptual frameworks, and decision models were proposed. Out of the ten articles, four of them are authored or co-authored by Dr. Dmitry Ivanov, who is excelling in the study of viability and resilience for SCs.

3.1. Keyword Co-Occurrence Analysis of the Entire Corpus

The keyword co-occurrence network, as shown in Figure 4, consisted of 204 items grouped into four clusters. Clusters are a set of closely related nodes and are not overlapped, i.e., one item cannot be in two clusters. The size of nodes is based on the number of their occurrences in the document. The clusters are named and numbered into four research categories, as shown in

Table 3. Cluster categories based on VOSviewer.

Cluster Name and Number	(1) Digitalization for Resilience and Sustainability	(2) Additive Manufacturing during the Pandemic	(3) Supply Chain Decision Models	(4) Food Supply Chain
	resilience, sustainability, analytics, artificial intelligence, industry 4.0, blockchain, digital, technology, innovation	3D, coronavirus, mask, manufacturing, protective, purchasing, virus, spread, pandemic	COVID-19, decision-making, distribution, modeling, optimization, programming, simulation, transportation, vaccine	agriculture, food, farming, market, production
Number of items	69	56	51	28

. It has been observed that cluster one is contributing the most in network with 69 items.

The overlay network visualization and the clusters show the four relevant SC research streams studied during the current pandemic: Digitalization for resilience and sustainability (red cluster), Additive manufacturing during the pandemic (green cluster), SC decision models (blue cluster), and Food SC (yellow cluster).

In

Table 4. The link and total link strength of the top ten occurrence keywords.

Rank	Keyword	Cluster Number	Occurrences	Links	Total Link Strength
1	supply	1	459	203	4974
2	chain	1	452	203	4901
3	COVID-19	3	416	203	4665
4	management	1	206	203	2520
5	pandemic	2	154	197	2022
6	analysis	4	113	197	1615
7	resilience	1	135	186	1504
8	sustainability	1	108	192	1362
9	system	3	85	198	1241
10	food	4	99	178	1227

, information regarding the top ten keywords is provided. The number of links is also shown, representing the co-occurrence connection between two keywords. The total link strength describes the number of publications in which two items appear simultaneously [43]. As expected, “supply chain” and “COVID-19” are the most occurring keywords per Table 4. Moreover, from Table 4, it can be observed that “resilience” and “sustainability” are the new research hotspot.

3.1.1. Digitalization for Resilience and Sustainability

During the pandemic, the analysis of SC resiliency has received increased attention, and its analysis has been performed from different perspectives considering various case studies, empirical research, and mathematical formulations [34,39,44,45]. Moreover, integrating resilience and sustainability has been considered a significant concern due to the changes in social implications and environmental practices in the time of disruptions [46]. Although SC resilience has demonstrated its fragility in the early pandemic days, it highlighted the potential transition toward sustainability [36,46]. The sustainability triple bottom line has been tested on a global scale. During the pandemic, economic and social (e.g., social distancing and remote work) sustainability efforts were observed. However, unintended environmental sustainability concerns occurred (e.g., excessive packaging for online deliveries, the usage of disposable masks, and the increase in last-mile deliveries). While there have been short-term sustainability gains, long-term effects are still to be discovered.

Furthermore, the fourth industrial revolution, often known as Industry 4.0, has enabled digital transformation, allowing digital connectedness and information development and sharing through different technologies (e.g., big data analytics, IoT, blockchain, etc.) [47]. Hence, industry 4.0 technologies have helped some SCs to become resilient to disruptions caused by the COVID-19 pandemic in various industries. Digital transformation could enable circular and sharing economies to boost the location of facilities onshore and shared resources. For instance, Spieske and Birkel [48] developed a framework between industry 4.0 and SC resilience. They applied their framework to an automotive company and found big data analytics could help increase SC resilience. Similarly, big data analytics also help MSMEs achieve sustainable competitive advantage [49]. Moreover, artificial intelligence (AI) can help improve SC resilience by ensuring last-mile delivery, giving individualized solutions to upstream and downstream SC stakeholders, minimizing the impact of disruption, and facilitating an agile procurement strategy [50].

Blockchain is another cutting-edge technology that guarantees the privacy, accuracy, and accessibility of all transactions and data. It is particularly safe for commercial operations since once the records are uploaded and cannot be changed without altering the earlier records [51]. Therefore, blockchain technology will allow transparency and data-sharing among the different stakeholders, which is especially important during disruptions, as normal operations are altered, and accurate information is needed to implement correct strategies. For instance, Ahmad, et al. [52] suggested a blockchain-based solution to automate forward SC operations and ensure data provenance about the COVID-19 medical equipment and their disposal. To promote collaboration, transparency, data integrity, and data provenance among stakeholders in the healthcare SC, Omar, et al. [53] developed a solution that combines blockchain and decentralized storage technologies. This promotes confidence and transparency among stakeholders by ensuring that only registered stakeholders are permitted to register and interact with the smart contract. While digitalization can help achieve SC resiliency and sustainability objectives, the SC ecosystem is heterogeneous. Therefore, roadmaps for their implementation need to be delineated according to the specific SC contexts.

3.1.2. Additive Manufacturing during the Pandemic

Millions of flights were canceled during the COVID-19 pandemic, which interrupted the delivery of various supplies, including protective gear to prevent virus transmission. Worldwide demand for medically related products grew during this time, although production using traditional industrial processes was hampered by lockdowns and transportation system disruptions.

Additive manufacturing (AM) is the process in which a product is created layer by layer and uses technology like computer-aided drawing to speed up production. AM was adopted for printing face shields and ventilators due to the surge in demand and because traditional manufacturing could not keep up with it. Although masks were not 3D printed since 3D printer materials are not healthy for long-term skin contact [54], other medical supplies were 3D printed. For example, Nazir, et al. [55] suggested using 3D printing to construct isolation wards, throat swabs, drugs with release control, and drones to deliver supplies to infected patients. Apart from reducing the time to produce products, equipment like PPE is complex and difficult to fabricate by traditional methods. Moreover, the customization of products can incur high costs if done by traditional manufacturing processes [56]. According to studies, it would be practical and affordable to produce reusable N95 respirators on a modest scale during a pandemic. Required quantities of reusables are less susceptible to SC disruption than disposables. With more research, these gadgets could replace disposable respirators in public health situations [57]. Although these advancements, the broad implementation of AM still faces many challenges, such as the investment required, the people’s expertise needed, and the definition of location for AM facilities.

3.1.3. Supply Chain Decision Models

SCs had to adapt to the new landscape during the pandemic. Therefore, the SC research focused on providing decision-making models (e.g., optimization, simulation) for helping to recover [58], continue working [59], and select suppliers [60] under the pandemic scenario. The decision models also were provided for organizing the distribution of protective gear like PPE, masks, face shields, and vaccines. One of the most significant underlying challenges was vaccinating people worldwide. Therefore, developing a robust vaccine supply chain (VSC) was needed to ensure the timely delivery of vaccines at the right place. Alam, et al. [61] categorized the VSC issues (e.g., a limited number of vaccine manufacturing companies, a long distance between vaccine stores and vaccination camps, a lack of vaccine monitoring bodies, etc.) and implemented an intuitionistic fuzzy decision-making trial and evaluation laboratory to rank the different issues. Considering the notions of equity and the requirements for cold, very cold, and ultra-cold SCs, Tavana, et al. [62] developed a mathematical programming model for equitable COVID-19 vaccine distribution in developing countries in the context of a location-inventory problem. Likewise, Gilani and Sahebi [63] proposed a data-driven robust optimization model to address the biased distribution of vaccines worldwide. They also implemented it in a real case scenario in Iran and found it less conservative than the classical model. During the pandemic, the distribution and sourcing of critical supplies were critical issues faced by SCs. While the developed models presented real and hypothetical scenarios, more collaboration with the different stakeholders is needed to streamline and transfer the solutions to concrete implementations.

3.1.4. Food Supply Chain

COVID-19 caused major disruptions among many SCs worldwide, and the food SC was also not immune to it. Agriculture is significantly impacted by the necessary COVID-19 prevention measures, with cascading short- and long-term effects. The resulting impacts can be used to understand the resilience of food systems worldwide [64]. Mobility restrictions and reduced purchasing power affected food security, impacting vulnerable groups like children dependent on school food programs and small farmers who cannot sell their produce at the markets [65]. For that reason, frameworks evaluating the performance of agri-food systems against COVID-19 disruptions have been explored in different countries [66,67]. Therefore, Kerr [68] suggested boosting international collaboration in the future to help supply networks continue to function during crises. While the COVID-19 issue has highlighted how dependent food security is on imports, this import dependency will result in further requests for self-sufficiency, protectionist policies, and reduced global engagement. It will partly depend on how economies change after the COVID-19 pandemic and which forces will rule international relations pertaining to agriculture and food.

3.2. Segmented Keyword Co-Occurrence Analysis

The co-occurrence analysis using keywords was performed on individual journals. The occurrence of the keywords was changed to five, owing to the database size per journal. The keywords were subject to lemmatization to ensure no repetitions took place for the analysis. The discussion in these sections consists of papers from the respective journals in our database.

3.2.1. Keyword Co-Occurrence Analysis for Sustainability

The network for the Sustainability journal consisted of 82 keywords which were grouped into 4 clusters. The co-occurrence network is shown in Figure 5, and the clusters are shown in

Table 5. Clusters—Sustainability based on VOSviewer.

Cluster 1 (Red)	Cluster 2 (Green)	Cluster 3 (Blue)	Cluster 4 (Yellow)
Manufacturing, business, logistics, performance, management	Resilience, sustainability, safety, risk, assessment	Agriculture, consumption, food, farming, coronavirus	Decision, strategic, technology, approach

.

The first cluster analyzes the different sectors affected by the COVID-19 pandemic and how their performance was evaluated and managed. In the current SC environment, businesses are more connected and more susceptible to being impacted by environmental changes like the pandemic due to the expansion of globalization. Hence, it is essential to have more robust supply networks that can effectively handle disruptions and unforeseen, unintended incidents [69]. For instance, Hilmola, et al. [70] surveyed Finland’s manufacturing and logistics companies on the early effects of the pandemic. Several companies mentioned that COVID-19 had increased their transportation costs moderately. A vast majority said they could meet customers’ requirements by collaborating, logistics routing, and enterprise resource planning (ERP). However, the implications of the second pandemic wave posed them to reevaluate their strategies. Therefore, evaluating and monitoring the SC performance when facing disruptions is needed. For example, Le and Nhieu [71] examined the negative impacts of the pandemic in Southeast Asia. They analyzed the different strategies and ranked them by creating a multi-scenario ranking tool to help managers with resource allocation decisions in post-pandemic productions. Similarly, Rodrigues, et al. [72] studied the weaknesses of SMEs and the importance of government measures to support these enterprises.

The second cluster discusses the importance of sustainable and resilient SCs. Bartle, et al. [73] highlighted the essential ways of improving environmental, financial, economic, and social sustainability. Wang, et al. [74] found a correlation between SC resilience and sustainability. Moreover, they highlighted that businesses must integrate the latest technological advances in their operations to ensure sustainability. For instance, the lockdowns caused many people to depend on food delivery services. Thus, Lin, et al. [75] suggested that a robust and scalable online food delivery service can help ensure the sustainability of this industry during disease outbreaks and other forms of disruption. In the same vein, a strategic process that includes collaborative network access, trust, and communication, can aid in identifying and managing collective risk, which could help to enhance SC resilience and sustainability [76]. Trabucco and De Giovanni [77] analyzed the uses of lean principles to improve SC sustainability. They found that lean omnichannel strategies and lean SC coordination could be used to ensure sustainability and that sustainability is linked to a firm’s probability of resilience.

The third cluster talks about the impacts on the agri-food SC. Food security has been a significant concern as it is one of the essential needs for human beings. Researchers have been studying the pandemic effects on food security in various countries to ensure that food insecurity is not repeated in the future. For instance, Rad, et al. [78] researched the impacts of COVID-19 on food security in Iran through a systematic literature review. They found that the pandemic and ecological constraints have caused direct and indirect damage to the agricultural sector. The pandemic has negatively impacted the food SC in Saudi Arabia by affecting demand and transportation [79]. Additionally, Paganini, et al. [80] surveyed small-scale farmers to analyze the impacts on food systems in rural areas of South Africa, Mozambique, and Indonesia. They suggested strategies for small-scale farmers to generate more income and build resilient food systems. Sitaker, et al. [81] propose Farm Fresh Food Box (F3B), a marketing approach that enables small-scale farmers to new markets at low risk. They initially found it to have low profits but helped farmers by enhancing their brand and increasing visibility. F3B also helped combat the pandemic’s impact on the food system by offering fresh goods with zero-contact curbside pickup at a convenient site.

The fourth cluster shows the tools involved in helping to mitigate the negative impacts of the pandemic on different sectors. Many researchers have proposed frameworks to tackle the problems faced in the pandemic and implemented their methodology on a case study to back their theory. Decision-making is an important aspect that can make or break a business during highly uncertain situations. Baležentis, et al. [82] proposed a multi-criteria decision-making framework for helping analyze decisions in improving resilience in agricultural SCs. They used a Pugh matrix approach for initializing the weights and performed a Monte Carlo simulation to ensure the results were accurate. Another major issue was the masks during the pandemic in terms of both demand and quality. Setiawan, et al. [83] proposed a closed-loop SC to ensure proper planning is done in terms of production and recycling. They aimed to maximize profit and minimize the environmental impacts. However, they found these both to be conflicting with each other. Since the advent of Industry 4.0, multiple technologies have been used to improve the operational aspects of businesses. Santoso, et al. [84] analyzed the benefits of adopting ERP technology and green SC management. They provide insights to managers to adopt ERP to enhance operational performance.

3.2.2. Keyword Co-Occurrence Analysis for IJLM

The IJLM network consisted of 15 keywords that were grouped into 3 clusters. The co-occurrence network is shown in Figure 6, and the clusters are shown in

Table 6. Clusters—IJLM based on VOSviewer.

Cluster 1 (Blue)	Cluster 2 (Red)	Cluster 3 (Green)
disruption, management, risk, resilience	supplier, sustainability, food	capability, COVID-19, logistics, dynamic

.

The first cluster mentions mostly papers related to risk management and disruption management. For instance, Ivanov [85] proposed the active usage of resilience assets (AURA) framework to gain lean resilience by reducing the effort for disruption prediction and the value creation from resilience assets. Dohale, et al. [86] analyzed the Indian apparel industries and the disruptions faced due to the pandemic. They adopted a Fuzzy Analytical Hierarchy Process (FAHP) to rank the apparel industry’s different supply chain risks (SCRs) and identify the severity of the SCRs. With a focus on logistics Service Providers (LSPS), Hohenstein [87] studied LSPS problems during the pandemic as many countries had closed borders abruptly, and their demand increased. Thus, LSPs had to deal with these problems swiftly.

The second cluster talks about the capabilities of logistics systems and the resilience observed by various industries. Logistics was a significant issue due to the changing market circumstances, and several researchers have addressed these hassles. Thus, Dovbischuk [88] surveyed 83 Logistic Service Providers (LSP) to understand their problems during the pandemic. A conceptual framework was used in that study to group various innovation-oriented capabilities (i.e., dynamic resilience, inter-firm resources, innovation, and organizational learning) for achieving more dynamic resilience. It was discovered that those capabilities were statistically significant for Logistics Service Quality and business performance. Additionally, the value of big data and AI in improving SC resilience capabilities have been explored [50,89].

The third cluster speaks about supplier selection and sustainability. Yassine [90] proposes a mathematical model to optimize the ordering quantity considering the fact of uncertainty in suppliers. Apart from uncertainty in the flow of supplies, it is also essential to consider different supplier auditing strategies during times like the pandemic. Remote supplier audits are a great way of removing travel and lodging expenses, thereby acting as a stepping stone for the future redesign of the supplier audit process [91]. This cluster also mentions how COVID-19 affected existing sustainable practices [92] and the importance of sustainability in a post-pandemic scenario [93].

3.2.3. Keyword Co-Occurrence Analysis for OMR

The number of papers from the OMR Journal in the database is significantly less than those published by the top contributor. Therefore, the co-occurrence network contains a single cluster, as shown in Figure 7. The main topic discussed in the journal is SC resilience during the pandemic. For instance, Das, et al. [94] proposed a multi-criteria decision approach to identify critical factors to provide strategies that could increase resilience. Likewise, Badhotiya, et al. [95] assessed Indian manufacturing firms by translating qualitative expert inputs to quantitative assessment using structural modeling and Bayesian networks. In another attempt to contain the pandemic, Chowdhury, et al. [96] studied the strategies to deal with the pandemic’s impacts on the food and beverage industry. In the same vein, Ivanov and Dolgui [45] developed a digital twin for stress testing of SC resilience.

4. Text Mining and Analytics Results

4.1. Title Analysis

For analyzing the titles, bigrams were used to understand the most frequent topics mentioned by researchers in their work. Title analysis quickly identifies the methods and expected results of a study. Studies have shown that titles that contain results and are short receive a higher number of citations [97,98]. This influences researchers to follow the same technique for writing their titles. Therefore, this analysis helps derive quick insights from the paper titles to understand the topics researchers in a particular domain must address.

The top 20 frequent bigrams are shown in Figure 8. From the analysis, the most frequent bigram is “Case Study,” which tells us that many researchers in this field have used a case study to apply their research methodology and support their arguments. Mainly case studies related to the food industry (e.g., [79,96,99]) and healthcare (e.g., [38,56,100,101]) were presented among other areas such as logistics, automotive industry, and tourism to analyze risk mitigation strategies (e.g., [86,102]). The pandemic has highlighted the importance of fulfilling basic needs. Therefore, food and healthcare are the most studied.

The next most frequent is “Literature Review,” which tells most of these papers have studied the state-of-the-art. For instance, reviews on healthcare operations and SCM [103], structured reviews on future trends of SCM [104], the impact of epidemic outbreaks on SCs [37], systematic reviews on SCs [105], and the intersection of SC and industry 4.0 [48] have been presented. Systematic literature reviews tend to be more popular than bibliometric reviews.

In order to provide solutions to the arising problems, “digital technologies” such as “blockchain” [106,107,108,109], “artificial intelligence” [50,110,111], and “big data” [112,113] have been explored. Moreover, uncertain conditions have encouraged implementing “robust optimization” [63,114].

The impacts of the pandemic have been assessed by the “firm performance” and “ripple effect.” SCs will continue looking for “resilient,” “sustainable” operations and even a “circular economy.” Moreover, “protective equipment” and “food systems and security” will be paramount when facing another pandemic.

4.2. Topic Modeling by Journals

The abstracts were passed through pre-text processing techniques, as shown in Figure 3, and subject to LDA, which generated four topics for each journal, presented in the following sub-sections.

4.2.1. Topic Modeling for Sustainability

The topics obtained after using LDA for topic modeling for the Sustainability journal are shown in

Table 7. Topic modeling—Sustainability.

Topic 1	Topic 2	Topic 3	Topic 4
blockchain, operation capability, sustainable procurement, factor food	automobile industry, panic buying, mask manufacturer, clothing company	circular economy, credit risk, industrial symbiosis, decision making	US beef, food aid, food security, food safety, food waste

. The first topic relates to using blockchain technology to help SCs recover from the pandemic. Research has shown that trust is an issue between partners that affects transactions as they need to provide information to improve visibility when facing disruptions [115]. For instance, Bekrar, Cadi, Todosijevic and Sarkis [107] discussed the concerns of adopting blockchain technology in transportation and reverse logistics. Moreover, Li, et al. [116] noted that SC integration and competitive performance are both positively impacted by blockchain operation capabilities. Therefore, they highlighted that SC integration has a significant mediating effect on this relationship. This way, blockchain presents a way of securely transferring information between parties. For example, using digital technologies (e.g., blockchain) can enhance sustainable procurement, as transparency allows for determining the criticality of components and tracking emissions and audits.

The second topic has papers that analyze the impacts on specific industries, such as the automotive, clothing, and mask industries. For instance, Eldem, et al. [117] studied the Turkish auto industry and presented a framework for improving operations when facing vulnerability, uncertainty, complexity, and ambiguity. Hence, they suggested to re-design resilient SC management by considering recovery plans. Chhimwal, et al. [118] analyzed the ripple effect (disruption risk propagation) using the Bayesian network method. They concentrated on an automobile organization’s circular SC network, demonstrating that circularity could be a source of risk propagation. The pandemic also affected the mask industry as the demand increased, and there was panic buying for them. However, mask manufacturers are vulnerable due to challenges such as the lack of monetary funds to increase capacity and the possibility of overcapacity [119]. Therefore, manufacturers and governments should collaborate. The optimal trajectory of mask manufacturing can be supported by the SC members’ investment efforts in production technology through the joint contract of two-way cost sharing and transfer payment [120].

Decision-making models and mitigation strategies employed are discussed in the third topic. For instance, decision-making in emergency healthcare SC poses different challenges, such as keeping corporate social responsibility while balancing profit [121]. Additionally, the potential of Industry 4.0 for business continuity and long-term sustainability has been studied to provide help in the decision-making process [122]. Moreover, the circular economy and other technological innovations were found to help businesses improve their situation [123]. Still, there is no evidence favoring the pandemic as a determinant of the circular economy followed by manufacturing companies [124].

The fourth topic deals with the various food SC issues addressed. Food SCs have been dramatically disrupted, causing food insecurity by directly affecting agriculture inputs, availability of labor, financial loss, and food price enhancements [78,125]. In terms of the US Beef industry, the farmers and consumers were affected due to the pandemic. It was found that customers were paying higher prices while farmers received less than the predicted amount. After further examination, it was found that the wholesale prices were adjusted more quickly than the retail or farm prices [126].

4.2.2. Topic Modeling for IJLM

The topics generated using the topic modeling approach for this journal are presented in

Table 8. Topic modelling—IJLM.

Topic 1	Topic 2	Topic 3	Topic 4
disruption management, freight transport, social sustainability, SDC view, AI adoption	food service, social capital, crisis management, mitigation strategy, apparel industry	buying firm, information sharing, perishable food, response epidemic, key strategy, leadership strategy	lead time, total cost, operational excellence, proposed model, technology related challenge,

. The first topic discusses social perspectives in different areas, such as coping with disruptions [127], the role of innovations for freight transport when facing crises [128], stakeholder dynamic capabilities (SDC), and the adoption of AI [129]. The second topic examines the implications for perishable SCs such as food [130] and the apparel industry [86] and studies crisis management strategies and the generation of social capital to improve SC operations [131]. The third topic explores the buyer-supplier relationship and the countermeasures taken to mitigate the impact of COVID-19 [132], the value of information sharing to improve SC visibility [133], and the implementation of leadership strategies [134]. The fourth topic studies the application of technology such as blockchain [109] and the analysis of the healthcare SC considering lead time and total cost [100] to achieve operational excellence.

4.2.3. Topic Modeling for OMR

Four discussion topics for the OMR journal were extracted, as shown in

Table 9. Topic modeling—OMR.

Topic 1	Topic 2	Topic 3	Topic 4
ripple effect, smart technology, SC crisis, SC crisis mitigation, crisis mitigation strategy, resilience	stakeholder engagement, tourism hospitality, i4.0	circular economy, recycling practice, remanufacturing recycling, digital technology	AI SCR, organizational performance, viability perspective, stress testing

. The first topic discusses how to build SC resiliency and mitigate the ripple effect when facing a crisis such as the current pandemic [94] and the strategies to deal with the impacts [96]. The second topic deals with the role of stakeholder engagement for resilient vaccine SC [135]. In contrast, parameters and drivers to build a resilient SC Tourism and hospitality [136] are also considered in this topic. The third topic considers sustainability and circularity perspectives such as recycling, reusing, remanufacturing, and repairing with the intersection of digital technology [137] to allow the continuity of such practices when facing the pandemic as nations look for alternate resource use. The fourth topic includes the consideration of an SC viable ecosystem to satisfy primary human needs [45] and the implementation of technology [111], as well as organizational aspects of the mitigation process [102].

4.3. Distinctive Discussed Topics Using TF-IDF among Journals

With this analysis, distinctive topics among journals were found. The higher the TF-IDF score, the more unique the word is to a particular document. In this context, the words that have the highest TF-IDF are the words that occur most frequently in a specific journal and least frequently in other journals. This could imply topics that have received particular attention in each journal.

Table 10. Distinctive words for the top-ten contributing journals.

Sustainability		International Journal of Logistics Management		Operations Management Research		Annals of Operations Research		International Journal of Production Research
Word	TF-IDF	Word	TF-IDF	Word	TF-IDF	Word	TF-IDF	Word	TF-IDF
PPE	0.7658	Transport	0.6417	FSCs	0.5060	SCV	0.6187	Mapping	0.7210
Mask	0.7153	SCR	0.6157	Circular	0.4678	VSC	0.6052	Masks	0.5428
Blockchain	0.6815	Audits	0.5858	DMs	0.4619	Vulnerability	0.5972	Multiplier	0.5294
Automobile	0.6261	ODFD	0.5540	Turbulent	0.4601	BCT	0.5465	Adaptation	0.5080
F3B	0.6142	AI	0.5241	Remanufacturing	0.4413	Procurement	0.5190	SDMs	0.4642
Credit	0.5781	OCs	0.5215	FGLS	0.4271	Recovery	0.5130	Enterprise	0.4092
Military	0.5727	PFSCs	0.4936	Ripple	0.3994	Concurrent	0.4865	Strategies	0.3735
TSC	0.5678	Riders	0.4847	Mitigation	0.3932	Vaccination	0.4859	Internal	0.3704
Air	0.5648	Fashion	0.4783	SCR	0.3932	Blood	0.4834	Information	0.3666
Satisfaction	0.5571	Drone	0.4763	Impacts	0.3760	Option	0.4619	Resilience	0.3512
Computers and Industrial Engineering		PLoS ONE		International Journal of Environmental Research and Public Health		International Journal of Logistics Research and Applications		Benchmarking
Word	TF-IDF	Word	TF-IDF	Word	TF-IDF	Word	TF-IDF	Word	TF-IDF
SME	0.5316	PPE	0.6054	Impacts	0.5716	Risks	0.6618	Mitigation	0.5859
SSS	0.4859	Vaccines	0.5528	Food	0.5548	AFSC	0.5473	SCF	0.5422
Measures	0.4348	Tokyo	0.5257	Global	0.4193	CE	0.5280	Hotel	0.5094
Industry	0.4341	Losses	0.4259	AV	0.3962	SCU	0.5074	SMEs	0.5023
Payment	0.4236	Channels	0.4155	Sharing	0.3851	Financing	0.4980	Global	0.4843
Cross-docking	0.3857	Items	0.3833	Commerce	0.3767	Enablers	0.4664	Restaurant	0.4840
Essential	0.3796	Ventilator	0.3672	Districts	0.3747	FSCs	0.4664	SCR	0.4734
Sharing	0.3731	Food	0.3481	Strategies	0.3674	Goals	0.4155	MSMEs	0.4671
Perishable	0.3696	Respirators	0.3420	Response	0.3653	Agricultural	0.3935	Tourism	0.4631
Compositional	0.3638	Marketing	0.3324	Products	0.3634	SSCs	0.3836	IoT	0.4596
Acronyms used in the Table
Acronym	Meaning			Acronym		Meaning
AFSC	Agri-Food Supply Chain			PFSC		Perishable Food Supply Chain
AI	Artificial Intelligence			PPE		Personal protective equipment
AV	Airborne Virus			SCF		Supply Chain Finance
BCT	Blockchain Technology			SCR		Supply Chain Risks
CE	Circular Economy			SCU		Supply Chain Vulnerability
DMs	Decision Makers			SCV		Supply Chain Viability
F3B	Farm Fresh Food Box			SDMs		Social Distancing Measures
FGLS	Feasible Generalized Least Squares			SME		Small and Medium-sized Enterprises
FSCs	Food Supply Chain			SSCs		Sustainable Supply Chains
IoT	Internet Of Things			SSS		Sustainable Supplier Selection
MSMEs	Micro, Small & Medium Enterprises			TSC		Tourism Supply Chain
OCs	Operational Challenges			VSC		Vaccine Supply Chain
ODFD	On-Demand Food Delivery

presents the uncommon words and TF-IDF scores for the top ten contributing journals.

For the Sustainability journal, “PPE,” “mask,” and “blockchain” are unique words. For the IJLM, “transport,” “SCR,” and “audits” are the most distinctive words. In the same manner, for the OMR journal, “FCSCs,” “circular,” and “DMs” are the most distinctive terms. The distinctive words for each journal represent future research avenues for those journals.

The words in Table 10 are color-coded to show the similarities among the different journals. For example, both Sustainability and Annals of Operations Research involved papers about blockchain technology (Green), as shown in their most distinctive words. The vaccine supply chain (Light Blue) was unique in Annals of Operations Research and PLoS ONE journals. The terms “PPE,” “Mask,” and “Ventilator” refers to the safety measures against COVID-19 (Yellow), which are the most distinctive in the Sustainability journal, International Journal of Production Research (IJPR), and PLoS ONE journal.

Supply chain resilience (Red) is the most distinctive word in the IJLM, the OMR journal, and the IJPR. “Perishable food supply chain,” “FSC,” “On-demand food delivery,” “F3B,” and “AFSC;” all relate to the food supply chain (Orange) and are distinctive in multiple journals like OMR, Sustainability, Computers and Industrial Engineering (CAIE), IJLM, International Journal of Logistics Research and Applications, and Benchmarking journals. “Enterprise,” “SME,” and “MSME” correspond to the COVID-19 effects on the smaller manufacturing industries (Dark Blue); these are discussed in IJPR, CAIE, and Benchmarking. Sustainability and Benchmarking discussed the tourism supply chain (Gold). Although food supply chains have been discussed across several journals, they are still many aspects to be investigated. Thus, representing an affluent area of study. While each journal has a specific scope, the research is explored from different perspectives during crises, which is observed from the diversity of topics discussed.

4.4. Distinct Topics of Discussion in the Entire Corpus

TF-IDF scores were obtained for the whole database to find the most prominent scores among all the abstracts. It is important to note that the TF-IDF score depends entirely on the given corpus and that the terms are not the most common in the database. Therefore, the analysis of the words represents opportunities for future research.

The overall top ten TF-IDF scores are presented in Figure 9. The obtained terms can be classified into three broad research streams: healthcare SC, food SC, and SC operations.

4.4.1. Healthcare SC

The most distinct term in the corpus is “Plasma.” Convalescent plasma therapy has been used to treat various viral diseases like the Machupo virus, Ebola virus, Junin virus, and more, which yielded positive results in reducing viral loads [138]. This same method was adopted during the COVID-19 pandemic to treat patients [139]; when an individual recovers from the virus infection, the plasma extracted from them is known as convalescent plasma [140]. During the different waves worldwide, the demand for plasma was surging beyond the supply limit [141]. Therefore, plasma SC models have been briefly discussed for facility location-allocation problems [140], scheduling [142], and dynamics of plasma demand [143]. To overcome this issue, it was essential to design an efficient model to facilitate the collection and distribution of plasma.

Another health-related term is “Cancer.” The COVID-19 pandemic’s effects on the manufacturing and service sectors have been widely discussed but also impacted other fields. The COVID-19 pandemic has halted much research due to the safety measures imposed during the initial stage, i.e., lockdown, quarantine, and other protocols. Therefore, the COVID-19 pandemic has impacted pharmaceutical SCs and priorities for cancer research due to the non-pharmaceutical methods taken to contain it (e.g., social distancing). This situation led to widespread healthcare deficiencies due to limited access to cancer care and decreased early cancer detection rates [144]. Consequently, the delays on the SC for cancer drugs would observe a delay.

Although PPE has been widely used during the pandemic, the word “Mask” is distinctive in the corpus. Masks were one of the most used PPE worn by healthcare workers and the public during the pandemic. The World Health Organization announced a scarcity of PPE by March 2020, putting healthcare workers worldwide in jeopardy. According to estimates, the industry would need to expand the production of medical masks by 40% if it wanted to prevent frontline personnel from being “dangerously ill-equipped to care for COVID-19 patients” [145]. Therefore, it was necessary to analyze different aspects of the mask SC to optimize the recovery strategies [146], tackle arising problems such the mask SC sustainability [147], and employ additive manufacturing to provide face masks [148].

4.4.2. Food SC

The term “Cattle” was the second-highest TF-IDF score in the corpus. The COVID-19 pandemic affects the supply and demand sides of different food ingredients. A significant disruption to the SC and loss of demand is caused, particularly by the closure or capacity reduction of the food service sector. The disruption could lead to an oversupply of fruits and vegetables and downward pressure on their pricing [149]. Çakır, Li and Yang [149] have shown that the US fresh produce SC has been resilient to the COVID-19 pandemic disruption, as there has been no statistically significant impact on wholesale prices. In contrast to the fresh produce industry, the cattle industry was estimated to face a 13.6 billion dollars loss, according to an economic damage report in April 2020 [150]. The Canadian beef SCs have been examined [151,152] according to customer changes, supply responses, and labor market constraints.

“Poultry” is another unique term in the database. Like the beef and cattle industry, the poultry industry faced SC disruptions due to the pandemic. The poultry industry is one of the major commercial sectors in the world’s agricultural commodities. According to research by the National Chicken Council, the United States supports 1,682,269 jobs, $441.15 billion in economic activity, and $34 billion in government income [99]. Initially, during the pandemic, the hospitality sector, like restaurants and hotels, had to face a lot of setbacks, which in turn caused a ripple effect on the poultry industry. The US poultry broiler industry was strong at the beginning of 2020. When the COVID-19 effects started, impressive production levels began to decline. The closure of restaurant dining rooms impacted consumer demand for chicken, while workforce interruptions impacted the SC for chicken. By April 25th, egg production was 8.3% lower than the previous week [153]. Yazdekhasti, Wang, Zhang and Ma [99] proposed a multi-modal stochastics SC model for the COVID pandemic.

4.4.3. SC Operations

“Reshoring” is the term with the third-highest TF-IDF score. Reshoring is moving production operations back to the parent company’s home nation, regardless of who owns the relocated operations [154]. The COVID-19 pandemic has caused several SC disruptions, and reshoring is a technique with excellent potential for mitigation. It helps improve customer satisfaction but also helps reduce the reliance on China for sourcing [155]. Although reshoring could alleviate disruption impacts, in-deep analysis per industry is needed to discover the strategic, tactical, and operational impacts.

Due to global SCs, maritime container shipping (MCS) is critical to delivering materials and goods to different regions. Since shipping containers were invented, the relevance and prevalence of maritime shipping increased as this kind of transportation became more affordable and effective [156]. Retail businesses depend significantly on maritime shipping for long-distance shipping. In order to handle their transportation logistics, retail companies frequently turn to third-party logistics firms [157]. Imbalances in trade brought on by the COVID-19 pandemic have produced blank sailings and the most unpredictably ever measured freight rates, making it vital for MCS businesses to develop resilience to deal with future occurrences of a similar nature [158]. Maritime transportation has been chaotic in recent years. While some seaports operate at total capacity, the efforts are not enough to accommodate the increasing demand. As the SCs have been exacerbated, the MCS requires a joint effort from third-party logistics to design strategies that sustain growth and the growing demand.

SC “viability,” characterized as the ability of the SC to sustain itself and endure in a changing environment through structural redesign and performance planning for long-term effect, has attracted attention [34]. Agility, resilience, and sustainability are three viewpoints on viability, regarded as SC’s core characteristics. SC viability can adjust to beneficial changes, take in harmful interference, recover from short- and long-term disruptions, and continue operating [159]. The COVID-19 problem profoundly impacts SCs in terms of volume, complexity, severity, and impact duration. It has caused a sudden and catastrophic transformation in the business environment and beyond [160]. The importance of SC viability has increased even more after the advent of the pandemic, as similar occurrences may happen again. Ruel, El Baz, Ivanov and Das [160] have found that the critical aspect of SCV is the constant reconfiguration of SC structures in an adaptive way to enable long-term survival.

To attain SC viability, “mapping” is required to gather information about suppliers and create a global map for the whole supply network. It helps identify the inherent risks in the SC and has many other benefits. By using SC mapping, companies can tell which areas would be affected by disruptions and try to mitigate the risks as soon as possible. For instance, Norwood and Peel [161] highlighted how companies could benefit by using SC mapping while highlighting the SC disruptions caused by the pandemic.

SCs operate in a complex ecosystem where various SCs intersect, allowing “Cross-industry Production (CIP),” which enables businesses to temporarily switch production lines during a crisis to make goods in industries they have never done before [162]. In 2020, there was great concern regarding the shortage in the SC of PPE [163]. In this scenario, an example of CIP would be garment industries with appropriate changes that could manufacture medical garments and keep the company running during a crisis. Although several industries possess the capabilities of producing a variety of products for their market, more research is needed to unveil how organizations can contribute and collaborate with other companies when facing disruptions. Moreover, considering how those organizations can evolve to satisfy different markets needs to be explored.

5. Discussion and Implications

The COVID-19 pandemic has highly impacted the SCs’ landscape. While many of the effects are still to be unfolded, after a year and a half of its start, the SC panorama has observed a shift in the mindset of SC practitioners, researchers, and even governments, to look for ways to better prepare for unexpected events. It is observed that the study of SC resiliency has attracted much attention. However, as SCs depend on different configurations, products, and systems, multifaced research is required to adapt accordingly to achieve SC resiliency. Hence, a sheer variety of studies have been presented. Moreover, from the review, it is noted that sustainability issues have been considered in parallel to the resiliency study. While economic sustainability is critical during crises, social and environmental sustainability needs to be considered. Working toward the design of a sustainable and resilient SC is still a challenge to be addressed. However, thanks to the new digital technologies, visibility among the different SC stakeholders could be attained. Though, there are still many opportunities to be explored to enable collaboration among the parties and gain trust for information sharing.

5.1. Theoretical Implications

This research delivers significant theoretical contributions to the state-of-the-art by conducting a literature review on research trends in SC strategies during the pandemic. The study was performed using quantitative methodologies utilizing bibliometric and text analytics. Unlike traditional literature reviews based on qualitative data and subject to bias, the approach undertaken enables the discovery of hidden patterns in the literature.

The implementation of the different methodologies allows for complementing the information provided by each method. For instance, the keyword co-occurrence network and title analysis unveiled general trend topics. Topic modeling offers more specific and hidden topics according to their probability. Moreover, TF-IDF uncovered topics not widely discussed that could not have been discovered otherwise. Those topics represent research opportunities to explore.

As the number of publications increases, literature surveys analyzing reference by reference become taxing and unmanageable. The proposed approach employed bibliometric and text analytics. The latter forming part of big data analytics. Therefore, implementing the different methodologies enabled the analysis of various data, such as titles, keywords, abstracts, and publishers, which allowed the discovery of frequent and hidden topics.

This work will be instrumental for researchers to observe at a glance the previous research and the trends studied in the scientific community and to provide future research agenda.

5.2. Practical Implications

This research provides practical implications for SC practitioners and researchers. The keyword co-occurrence network for the whole corpus provided insights into the research discussed in the domain. Due to the number of articles published, the clustering analysis allowed the identification of research streams that have been widely discussed. Moreover, the title analysis quickly permitted the observation of the most frequent tools and analysis techniques in the literature.

For SC researchers, information is provided regarding the top publications, research trends, and journal perspectives. From the Sustainability journal keyword co-occurrence network, the journal’s broader scope can be observed. The main three of the four clusters identified in the whole corpus are presented in this journal, too (i.e., agri-food SCs, resilience and sustainability, and SC decision models). As the top contributing journal, its research influence is observed. Similarly, the results from the topic modeling for this journal are similar. However, the latter results tend to be more specific. Regarding IJLM, the keyword network showed that its scope concentrates more on studying resilience for logistic systems. However, the topic modeling for that journal presented more details and specific topics. Concerning the OMR journal, the keyword network displayed only one cluster related to SC resiliency, which is reasonable due to the number of articles published in that journal being significantly fewer. Nevertheless, it can be noted that the topic modeling methodology can uncover more topics. It is relevant to highlight that co-occurrence networks were created with only keywords, while the topic modeling methodology employed the abstracts of the articles. Therefore, the keyword co-occurrence network and topic modeling methodologies can be used together to complement the information obtained and provide a general picture of the keyword network and details with the topic modeling.

Regarding the TF-IDF scores, the segmented analysis showed which topics have been scarcely studied in each journal. For instance, “PPE,” “Transport,” and “FSCs” are the unique words for the top three contributing journals, respectively. Therefore, topics related to those keywords represent opportunities to study in more detail and have a greater chance to be published in those outlets, as not many articles have explored those topics in those journals. Concerning TF-IDF scores for the whole corpus, the words obtained are the most distinctive, meaning that just a small fraction of articles have discussed them. Hence, they represent opportunities for the scientific community to continue exploring them.

For SC practitioners, this research can provide valuable insights and help them to understand the SC panorama and the advancements. The research findings support the need for more information sharing, transparency, and visibility to enable stakeholder collaboration. Thus, SC practitioners can call top managers to engage in the implementing and democratizing digital technologies at all organizational levels. Moreover, resilience and sustainability need to be examined for long-term implementation, considering not only economic but social and environmental factors. Implementing a circular economy, shared resources and the location of AM facilities is needed.

Governments and SC practitioners need to provide measures that allow SMEs to thrive in uncertain environments. Food SC managers and governments need to work together to establish policies that ensure the food supply is delivered during crises.

6. Future Research Avenues

The analysis of the co-occurrence keyword network, in combination with the text analytics tools, has shed light on relevant issues during the pandemic. On the other hand, TF-IDF analysis proposed some relevant research streams to be studied. Besides those topics, further investigation is required for critical issues as follows.

6.1. Cross-Industry and Intertwining of SCs

Cross-industry production was an essential strategy for business to continue their activities despite the impacts of the pandemic. Intertwined supply network includes entire groups of linked SCs that, when functioning correctly, ensure that society and markets are supplied with commodities and services [34]. The need for more research on the viability of intertwined SC has increased due to the pandemic and is comparatively new to the SC literature. Models that can describe and optimize their operations are required.

6.2. Shipping Industry

Water transport is one of the most common modes of transportation used by global SCs for cars, apparel, and other products. The pandemic has significantly affected MCS firms. As MCS firms face uncertainty, it is also essential to find solutions to handle the uncertain SC environment [158]. Therefore, a more in-deep analysis of MCS to improve SC resilience and its effect on sustainability requires further investigation.

6.3. Ripple Effect

The propagation of disruption from the original disruption place is known as the ripple effect, which can disrupt the whole SC. The pandemic unleashed ripple effects across many industries, especially global SCs [164]. Although the literature on ripple effect analysis is available, studies that analyze it in the pandemic scenario are limited. Thus, this represents an opportunity to explore.

6.4. Visibility and Optimization for PPE

PPE, such as masks, was essential during the pandemic. However, the SCs for masks faced many challenges to accommodate the increase in demand. Lack of visibility was proven to be one of the reasons for the shortage of PPE during the pandemic [165]. This shows that future researchers can work more to improve the literature in this field by helping SC managers to decide when to ramp up and down their capacities. However, such a decision could be challenging without the information on the whole system. Therefore, improving SC visibility increases the need for better technologies.

6.5. VSC Optimization

Although several models are available for optimizing the COVID-19 vaccine, more models for specific settings are needed (e.g., different vaccine types, populations, facility availability, etc.). Moreover, collaboration with governments to implement such models is required to ensure vaccine distribution equity. Consequently, the benefits can be observed in the population.

6.6. Tourism SC

One of the major industries affected was the tourism and hospitality sector [166], but there has not been significant research on how the tourism SC can overcome the pandemic challenges. Tasnim, Shareef, Dwivedi, Kumar, Kumar, Malik and Raman [136] have discussed SC drivers that could help build a resilient tourism SC. However, more studies are necessary to address customer behaviors during crises and the impacts at the operational level.

6.7. SMEs and SC Impacts

Many organizations of all sizes were affected by the pandemic. However, the analysis of the impacts is mainly concentrated on large-size organizations. Less than 5% of the reviewed literature addressed the effects on SMEs. However, SMEs could face even more significant challenges compared with large-size organizations. Thus, analyzing SMEs’ performance and implementing strategies is essential to help organizations thrive in uncertain SC environments.

6.8. SC Resilience and Sustainability

Resilience and sustainability have gained significant importance since the pandemic and have been critical for businesses to attain resilience to disruptions like the pandemic. Studies on resilience [167] in different industries and integration with sustainability have also been done. However, resiliency and sustainability concepts are multidimensional, and each industry has its limitations in implementing them. Therefore, the intersection of these two domains calls for a holistic analysis of different sectors.

6.9. SC Facility Relocation

The pandemic highlighted SC vulnerabilities, especially for globally distributed SCs. Hence, several organizations are looking to onshore, nearshore, and reshore their facilities. Tools to guide such decisions need to be further designed. Additionally, the effect of such strategic decisions needs to be explored in the long term to analyze their benefits.

6.10. Digital Technologies to Improve SC Resilience

Digital technologies (e.g., big data analytics, IoT, blockchain, digital twins, etc.) have demonstrated the value of information. While many articles have explored different technologies, their orchestration to create a smart SC ecosystem is still needed. Digital SC twins’ frameworks have been proposed. However, the broad implementation of them is still required for many industries. The design of cognitive SCs based on digital technologies needs to be explored to allow preparedness and recovery for unexpected events [168].

7. Conclusions

This paper identified significant contributions to the field of SC disruption during COVID-19, and insights for future research avenues were provided based on the bibliometric and text mining approach. The data was collected from the Scopus database consisting of 574 papers. According to the number of articles published per journal, the Sustainability journal was the highest contributor, followed by IJLM and OMR.

A co-occurrence network was visualized using the articles’ keywords to identify the discussed research themes. Four research themes emerged in the current database for the whole corpus: Digitalization for resilience and sustainability, Additive Manufacturing, Supply Chain Decision Models and Food Supply Chain. It was found that resiliency and sustainability are the research hotspots.

Moreover, papers’ titles were analyzed using bigrams, a rapid way to identify the SC tools used by various researchers. Consequently, it was found that most researchers use a case study to help prove their research. Case studies help provide a holistic review of a particular subject.

Additionally, individual keyword co-occurrence networks were analyzed for the top three contributing journals. The networks for each journal provided a broader idea of topics discussed in each journal. Moreover, LDA topic modeling was used to identify the topics in the top three contributing journals. The topics provided by the latter method showed to be more detailed than the results from the keyword networks. Hence, the LDA method provided a quick way of classifying papers into topics, making the review process easy and holistic.

Furthermore, using the paper abstracts, TF-IDF was used to find the unique topics among the top 10 journals. Among the distinctive topics, agri-food SC was studied by several journals. TF-IDF was also used on the entire corpus to identify unique issues of discussion that might require more attention due to their importance. Therefore, an in-depth analysis of the special topics to understand their significance is needed. For example, “Plasma” was found to have the highest TF-IDF score. Convalescent plasma was used to treat COVID-19 patients, so there was an increased need to collect and distribute the plasma efficiently through its SC. Thus, further research is required to optimize the plasma SC.

The topics highlighted by the TF-IDF scores represent research opportunities to be explored. Additionally, based on the full review, future research avenues were suggested for researchers to analyze further and strengthen the literature.

This research would help find the different problems faced by SCs during the pandemic and strategies used to combat it in a scenario where history would repeat itself. Performing a literature review manually by scanning hundreds of papers is an exhaustive and time-consuming task. Hence, this paper has used various tools, such as topic modeling and TF-IDF scores, to tackle this problem. The different methodologies implemented complemented each other and drew a big picture without missing critical details.

Like every study, this literature review has its limitations. Firstly, the existing literature is being updated as this paper is drafted, and this study only considers the papers collected during the initial search. Secondly, we utilize the LDA method for topic modeling. However, there are other methods for topic modeling which might lead to different results. The results can also vary regarding keywords used while performing the search and the custom stop words used during the data cleaning process.