A Multi-Stage T-SF MCDM Framework for Assessing Vulnerabilities in Cross-Border Logistics Chains

Abstract

The rapid expansion of cross-border e-commerce (CBEC) has reshaped global trade while exposing supply chains to increasing vulnerabilities, including infrastructure bottlenecks, regulatory fragmentation, and operational disruptions. This study develops an integrated framework to evaluate and enhance the resilience of Vietnam's cross-border e-commerce supply chain (CBECSC). A hybrid T-spherical fuzzy (T-SF) multi-criteria decision-making framework is applied in three stages. First, the T-SF Delphi method (T-SF Delphi) is used to identify critical resilience factors. Second, the T-SF Decision-Making Trial and Evaluation Laboratory (T-SF DEMATEL) method is employed to analyze causal relationships among these factors. Third, strategic interventions are prioritized using the T-SF Combined Compromise Solution (T-SF COCOSO) method. To ensure methodological robustness, the ranking results obtained from T-SF COCOSO are further validated through comparative analysis with the T-SF Measurement Alternatives and Ranking according to Compromise Solution (T-SF MARCOS) and the T-SF Technique for Order Preference by Similarity to Ideal Solution (T-SF TOPSIS). The findings indicate that Customize Products, Mutual Contingency Plan, Trade Policies, Safety Stock Inventory, and Reconfigure Contingency Strategy exert the strongest causal influence on supply chain resilience. Although the empirical analysis focuses on Vietnam, the proposed framework offers a transferable methodology for other emerging economies facing similar CBEC challenges.

Keywords

cross-border e-commerce supply chain resilience T-spherical fuzzy sets MCDM cross-border logistics

1 Introduction

Assessing the robustness of supply networks is critical, especially in today's turbulent environment. Many researchers have argued that disasters such as earthquakes, political unrest, fuel shortages, epidemics, and terrorism (Chen et al., 2013; Sawik, 2013) severely disrupt businesses’ ability to produce and distribute goods. The COVID-19 epidemic has raised awareness among academics and practitioners about the importance of establishing more resilient supply chains to reduce the potentially devastating impact of disruptions. Although there are numerous definitions of supply chain resilience (SCRE), they all agree on one fundamental point: SCRE refers to the system's ability to prepare for, respond to, adapt to, and recover from interruptions in order to maintain or restore normal operations at a targeted level of efficiency, while also improving the ability to adapt flexibly to changing conditions (Tukamuhabwa et al., 2015).

As economic globalization advances, cross-border e-commerce (CBEC) is emerging as a new mode of international trade development (Giuffrida et al., 2017; Liu et al., 2021). CBEC is a technique for selling products to consumers in other countries via online channels, such as websites or e-commerce platforms (Giuffrida et al., 2017). CBEC, like traditional e-commerce, offers a range of products, including both real and digital goods, as well as intangible services. The distinction is that CBEC can connect buyers and customers globally (Liu et al., 2021). A cross-border e-commerce supply chain (CBECSC) can be defined as an organizational structure led by core enterprises, designed to meet the needs of CBEC consumers through collaboration with supply chain partners, encompassing the entire process of product design, procurement, manufacturing, sales, and service (Liu et al., 2021). The CBECSC has a more complex structure than a traditional supply chain, as it encompasses foreign transactions and logistics.

CBECSC resilience has emerged as a significant challenge in global commerce, particularly in the context of crises such as the Russia-Ukraine conflict and the COVID-19 pandemic. The pandemic has increased global travel, distribution, and production, while the downturn in China—the world's largest industrial hub—has stressed supply chains, resulting in shortages, rising transportation costs, and inconsistent raw material supplies (Das et al., 2022). The conflict between Russia and Ukraine has impacted the production and trade of wheat, sunflower oil, corn, natural gas, oil, and fertilizer. Sanctions and trade restrictions have tripled or quadrupled food shortages and fuel costs, particularly in nations that export agricultural and petroleum products from the region, leading to global economic instability and concerns about food security (Zhou et al., 2023). Scholars from around the world have been closely following this topic. Guo et al. (2023) investigated the impact of COVID-19 on SCRE in agricultural e-commerce in China, specifically CBECSC resilience. Suali et al. (2024) investigated the importance of digital platforms in improving the resilience of e-commerce food supply chains to external interruptions, particularly in the context of the COVID-19 pandemic, which significantly impacted the food supply chain in the United Kingdom. Zaman et al. (2023) investigated the impact of digitalization on the service supply chain robustness of the banking industry in the e-commerce sector, utilizing a technology adoption model. Pu et al. (2024) developed a multi-period stochastic programming model to study a technique for optimizing the resilience of e-commerce supply chains based on geographical connection in the setting of disruption risk.

According to the Vietnam E-commerce White Paper 2023, 78.6% of the population used the internet, with 61 million people (78%) engaging in e-commerce and 28% buying products from foreign websites (Ministry of Industry and Trade of Vietnam, 2023). E-commerce, especially in Vietnam, is closely tied to CBEC. CBECSC faces external dangers, including pandemics, wars, plagues, and financial panics. These hazards disrupt and impair supply chains (Liu et al., 2021). Furthermore, Vietnam's logistics infrastructure has a unique positioning among emerging economies. While leading digital economies such as China have developed highly sophisticated CBEC logistics systems supported by overseas warehouses, dedicated cargo networks, and advanced digital technologies (Wang et al., 2023; Yan et al., 2023), many developing economies continue to face significant infrastructure bottlenecks, heavy reliance on single transport modes, and vulnerability to external disruptions (Banomyong et al., 2025; Gupta & Gupta, 2026; López et al., 2022; Nabil et al., 2024). Vietnam represents a transitional context, combining rapid e-commerce expansion with persistent logistics fragmentation and capacity constraints. Consequently, developing a resilience assessment framework in this environment provides a methodological approach that can potentially be adapted to both more advanced digital ecosystems and resource-constrained logistics systems across other emerging markets. This enables businesses operating in multiple countries to remain competitive and achieve success. From various perspectives, researchers worldwide have examined the resilience capabilities of CBECSC and drawn valuable conclusions. To the best of our knowledge, no research has been conducted on the resilience of the Vietnamese CBECSC. This study examines the primary factors that influence the resilience of CBECSC and its correlation with the suggested development initiatives in Vietnam.

CBECSC resilience is classified as a multi-criteria problem because it involves numerous dimensions and stakeholders. Various studies have used the Multi-Criteria Decision-Making (MCDM) technique to assess the factors that influence CBECSC. For example, Liu et al. (2021) employed Fuzzy Decision-Making Trial and Evaluation Laboratory (DEMATEL) and Interpretive Structural Modeling (ISM) Zhang et al. (2023), applied Fuzzy Analytical Hierarchy Process (AHP) and Fuzzy Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) techniques, and Nabil et al. (2024) created a hybrid model that included Simple Additive Weighting (SAW) and ISM. However, the standard MCDM technique fails to capture ambiguous and unclear information in expert opinions. As a result, research often combines fuzzy and MCDM approaches to address the constraints listed above. However, in addition to the robust development of fuzzy sets, recently developed Fuzzy Sets such as Spherical Fuzzy Sets (SFS) and T-Spherical Fuzzy Sets (T-SFS) have drawn significant attention from researchers due to their superiority over older Fuzzy Sets. SFS and T-SF are considered capable of precisely capturing information relevant to complex challenges like CBECSC resilience, as they represent information in a 3D spherical space (Anjum et al., 2024; Aytekin et al., 2024; Ju et al., 2021). Therefore, the TFS is the most suitable fuzzy set for addressing the uncertainties in this scenario.

The recent integration of MCDM frameworks with T-Spherical Fuzzy (T-SF) methodologies has addressed substantial research deficiencies (Aytekin et al., 2024; Görçün et al., 2024; Ju et al., 2021). The fuzzy and spherical fuzzy methods are ineffective against the T-SFS. This facilitates the exchange of information regarding expert opinions and uncertainty in intricate systems such as CBECSC. The new merging of T-SF approaches with MCDM methodologies fills significant research gaps. The T-SFS add reluctance to fuzzy and spherical fuzzy techniques, giving a nuanced foundation for describing uncertainty and expert opinion in complex systems like CBECSC. Traditional MCDM methods, such as TOPSIS and DEMATEL, use crisp or Triangular Fuzzy Sets, which oversimplify uncertainties. However, T-SFS improve decision-making granularity and accuracy. This study attempted to address two significant research deficiencies. Despite Vietnam's robust e-commerce growth, the majority of e-commerce firms are small to medium-sized, necessitating separate research to formulate suitable resilience development plans in comparison to other nations. This study integrates T-SF, a contemporary fuzzy set, with MCDM to examine the resilience of CBECSC in Vietnam. In particular, three advanced T-SF MCDM methods are applied in this research: T-SF Delphi to identify and validate critical resilience factors, T-SF DEMATEL to analyze causal relationships and classify factors into causative and affected groups, and T-SF Combined Compromise Solution (COCOSO) to prioritize resilience strategies using a multi-criteria framework. Reliability is ensured by comparing the findings to T-SF TOPSIS and T-SF Measurement of Alternatives and Ranking according to Compromise Solution (MARCOS), thereby addressing a methodological gap where other approaches are rarely employed for verification. This study demonstrates that T-SF methods can effectively handle multi-dimensional uncertainties and interdependencies, outperforming the AHP, Fuzzy AHP, and crisp DEMATEL in terms of uncertainty representation, causal analysis, and strategy prioritization. This paper pioneers the application of T-SF approaches to CBECSC resilience, creating a framework that overcomes existing limits.

This study aims to address this deficiency by examining the essential components that affect the resilience of Vietnam's CBECSC and their interconnections. The research aims are articulated through the subsequent inquiries:

RQ1 What are the key factors influencing the resilience of CBECSC in Vietnam?

RQ2 How do these factors interact within the resilience system, and which of them function as driving factors and which represent dependent outcomes?

RQ3 What is the prioritized sequence of strategies to strengthen the resilience of CBECSC in Vietnam?

This study employs an innovative T-SF-based MCDM framework to address these issues. Initially, we identify and validate critical CBECSC resilience variables with the T-SF Delphi methodology. Second, T-SF DEMATEL examines the connections between these parts. The T-SF COCOSO method finally ranks techniques that help people become more resilient. The research evaluates these results against T-SF TOPSIS and T-SF MARCOS for robustness. The contributions of this study are listed as follows:

(i)
First, it contributes to the scholarly understanding of CBECSC resilience by methodically identifying and organizing the important operational, institutional, and logistical aspects that determine vulnerability and resilience in the Vietnamese context.
(ii)
Secondly, the study methodologically contributes by showcasing the efficacy of modern T-spherical fuzzy multi-criteria decision-making (T-SF MCDM) techniques in analyzing intricate and interrelated logistics systems.
(iii)
Third, from a practical perspective, the study provides policymakers and logistics practitioners with a structured framework for prioritizing resilience-enhancing strategies. The findings support more effective risk mitigation, strategic planning, and coordination in Vietnam's CBEC logistics systems, while offering insights applicable to similar emerging-market contexts.

The structure of the rest of the paper is as follows: Section 2 reviews the relevant literature and identifies research gaps. Section 3 covers the T-SF approach and procedures. A Vietnam CBEC supply chain case study is used to demonstrate and discuss the results in Section 4. Section 5 summarizes the findings, discusses the limitations, and proposes future research directions.
2 Literature Review

2.1 Literature Review Procedure Based on Web of Science and Scopus

To ensure methodological rigor and comprehensive academic coverage, this study conducted a systematic literature review using two internationally recognized databases: Web of Science (WoS Core Collection) and Scopus. These databases were selected due to their extensive coverage in the fields of supply chain management, logistics, CBEC, and MCDM.

The search strategy was developed around the central concept of SCR, combined with contextual terms related to CBEC and international supply chains. The search strings were structured as follows:

Web of Science (TS – Topic: Title, Abstract, Keywords): TS = (“supply chain resilience”) AND TS = (“cross-border e-commerce” OR “e-commerce supply chain” OR “international supply chain”)

Scopus (TITLE-ABS-KEY): TITLE-ABS-KEY(“supply chain resilience”) AND TITLE-ABS-KEY(“cross-border e-commerce” OR “e-commerce supply chain” OR “international supply chain”)

This strategy ensured that the selected studies explicitly addressed both (i) SCRE and (ii) the context of CBECSC. No publication year restrictions were applied during the initial search phase to avoid excluding foundational studies. Search results from both databases were exported in CSV/Excel format, including the following metadata fields: title, authors, abstract, keywords, publication year, and DOI. The data processing procedure consisted of three main steps:

Step 1: Data Merging and Duplicate Removal.

The initial search yielded 57 records from Web of Science and 56 records from Scopus. These records were merged into a single dataset. Duplicate entries were primarily identified and removed based on DOI. In cases where DOI was unavailable, duplication was determined based on identical titles. After this step, 65 unique studies remained.

Step 2: Abstract Screening.

The 65 studies were screened based on their abstracts to assess their relevance to the objective of constructing a structured system of resilience factors in the context of CBECSC. At this stage, studies were excluded if they focused exclusively on technical optimization models without proposing a structured set of resilience factors, consisted primarily of policy commentaries lacking a clear conceptual or analytical framework, or examined performance outcomes without explicitly identifying resilience capabilities or structural constructs. After this screening process, approximately 20 studies were retained for full-text review.

Step 3: Full-Text Screening.

The 20 studies selected from Step 2 underwent a detailed full-text review. Studies were retained as core studies if they made substantial contributions to the structural understanding of supply chain resilience in cross-border or e-commerce contexts.

From this process, 11 core studies were retained and are presented in detail in Table 1. The remaining studies, although not explicitly proposing structured resilience factors, were nevertheless incorporated to reinforce the theoretical foundation of the study, provide contextual insights, and support the relevance of the identified resilience dimensions within the CBEC environment.

Table 1.
Core Studies on CBECSC Resilience.

Study Context Method Conceptual Foundation Resilience Structure Role

Liu et al. (2021) CBECSC Fuzzy DEMATEL + ISM Supply chain resilience perspective integrating Information Processing Theory (efficiency), High Reliability Theory (adaptability), and Social Exchange Theory (evolution capability) 3 primary capabilities (Supply Chain Efficiency, Adaptation, Evolution) → 12 secondary dimensions → 36 factors; outputs: cause–effect classification, centrality ranking, 4-layer hierarchical model Foundational structural backbone:

(i) Provides CBEC-specific 3-capability architecture.

(ii) Offers validated 36-factor taxonom.

(iii) Demonstrates causal interaction mapping among resilience drivers.

(iv) Legitimizes multi-level structuring before weighting.

(v) Methodological precedent for integrated fuzzy causal modelling prior to advanced T-SF-MCDM extension.

Nabil et al. (2024) Transnational e-commerce Entropy Weight Method (EWM) + SAW + ISM + SWOT Supply chain resilience theory and MCDM perspective 3 primary dimensions (Adaptability, Efficiency, Evolution) → 36 resilience factors → hierarchical ISM structure and strategic prioritization Factor taxonomy and structural validation:

(i) supports the multidimensional resilience framework.

(ii) provides empirically validated factor definitions.

(iii) offers hierarchical structuring logic for resilience drivers.

(iv) supports the classification of factors across efficiency, evolution, and adaptability capabilities.

Zhang et al. (2023) CBECSC Expert evaluation, Fuzzy AHP, Fuzzy TOPSIS, case study Balance State of Resilience and Vulnerability (BSRV); resilience–vulnerability trade-off logic SCR drivers (agility, visibility, collaboration, information sharing, risk sharing) and SC Vulnerability drivers (supply, operational, demand, environmental risk) → resilience–vulnerability equilibrium Conceptual balancing perspective:

(i) clarifies the coexistence of resilience and vulnerability.

(ii) supports trade-off reasoning in resilience strategy formulation.

(iii) reinforces the need for prioritization of resilience drivers rather than unconditional maximization.

Shan et al. (2023) E-commerce platforms SEM Behaviour–Capability– Performance (BCP) paradigm Supply chain integration (supplier, internal, customer) → supply chain resilience (risk management, agility, robustness, restoration) → sustainable performance (economic, environmental, social) Causal validation stream:

(i) empirically confirms integration → resilience → performance relationship.

(ii) legitimizes resilience as mediating capability.

(iii) supports importance of collaboration and information-sharing factors.

Russo et al. (2022) Retail supply chains Scenario-based experiments (ANCOVA; PROCESS macro) Procedural Justice Theory (PJT) + Interactional justice Service recovery resilience → Procedural justice → Customer satisfaction → Customer loyalty (moderated by interactional justice) Customer-centric resilience perspective:

(i) demonstrates behavioural outcomes of recovery capability.

(ii) highlights communication and fairness as moderators of resilience outcomes.

(iii) extends resilience analysis to service recovery contexts.

Wang et al. (2023) E-commerce supply chains Empirical / case-based Information Processing Theory + Organizational Ambidexterity Information-processing fit → exploration/exploitation balance → agility, redundancy, flexibility capabilities Capability development mechanism:

(i) explains how information-processing capability shapes resilience.

(ii) supports inclusion of agility, redundancy, and flexibility as key SCR capabilities.

(iii) provides theoretical grounding for interaction among resilience drivers.

Yan et al. (2023) CBEC Case study Information Processing Theory Information processing fit → 3 collaborative innovation modes (technology, management, market) →resilience capabilities (agility, redundancy) Collaboration-driven resilience mechanism:

(i) supports the role of knowledge exchange and digital collaboration.

(ii) provides causal logic linking collaboration to resilience capabilities.

(iii) reinforces upstream drivers in resilience systems.

López et al. (2022) Global fashion supply chain Calibrated Fuzzy Best–Worst Method; expert judgement; sensitivity analysis Decision-analytic resilience evaluation using fuzzy MCDM Two-tier resilience evaluation framework: resilience capabilities and performance indicators; calibrated weighting scheme Methodological benchmark:

(i) validates fuzzy expert-based weighting of resilience drivers.

(ii) supports methodological legitimacy of advanced fuzzy MCDM approaches.

(iii) provides comparative baseline for T-SF-based prioritization.

Banomyong et al. (2025) Cross-border supply chains Cross-Border Resilience Performance Index Supply chain resilience capability perspective integrated with Logistics Performance Index concepts Strategic capabilities: integrated flexibility & redundancy; integrated collaboration & agility → resilience performance index Measurement benchmark:

(i) operationalizes resilience into measurable capability indicators.

(ii) supports factor-based resilience evaluation.

(iii) provides reference for quantitative resilience assessment frameworks.

Ma et al. (2025) Cross-border logistics System dynamics modelling (causal loops, stock–flow simulation) Systems thinking and the feedback-loop resilience perspective Dynamic interactions among inventory, transportation delay, capacity adjustment, demand fluctuation, and recovery rate Dynamic interaction justification:

(i) demonstrates non-linear relationships among resilience drivers.

(ii) supports interaction-aware modelling.

(iii) reinforces the need for system-level resilience analysis beyond static factor evaluation.

Garola et al. (2023) Express logistics during COVID-19 Multiple case studies Resource-Based View and Dynamic Capabilities perspective Resource reconfiguration under disruption; key enablers: visibility, communication, digital technologies, prior knowledge Operational resilience evidence:

(i) validates micro-foundations of resilience in logistics networks.

(ii) highlights digitalization and communication as resilience enablers.

(iii) supports practical implications for logistics coordination and disruption response.

Study	Context	Method	Conceptual Foundation	Resilience Structure	Role
Liu et al. (2021)	CBECSC	Fuzzy DEMATEL + ISM	Supply chain resilience perspective integrating Information Processing Theory (efficiency), High Reliability Theory (adaptability), and Social Exchange Theory (evolution capability)	3 primary capabilities (Supply Chain Efficiency, Adaptation, Evolution) → 12 secondary dimensions → 36 factors; outputs: cause–effect classification, centrality ranking, 4-layer hierarchical model	Foundational structural backbone:
(i) Provides CBEC-specific 3-capability architecture.
(ii) Offers validated 36-factor taxonom.
(iii) Demonstrates causal interaction mapping among resilience drivers.
(iv) Legitimizes multi-level structuring before weighting.
(v) Methodological precedent for integrated fuzzy causal modelling prior to advanced T-SF-MCDM extension.
Nabil et al. (2024)	Transnational e-commerce	Entropy Weight Method (EWM) + SAW + ISM + SWOT	Supply chain resilience theory and MCDM perspective	3 primary dimensions (Adaptability, Efficiency, Evolution) → 36 resilience factors → hierarchical ISM structure and strategic prioritization	Factor taxonomy and structural validation:
(i) supports the multidimensional resilience framework.
(ii) provides empirically validated factor definitions.
(iii) offers hierarchical structuring logic for resilience drivers.
(iv) supports the classification of factors across efficiency, evolution, and adaptability capabilities.
Zhang et al. (2023)	CBECSC	Expert evaluation, Fuzzy AHP, Fuzzy TOPSIS, case study	Balance State of Resilience and Vulnerability (BSRV); resilience–vulnerability trade-off logic	SCR drivers (agility, visibility, collaboration, information sharing, risk sharing) and SC Vulnerability drivers (supply, operational, demand, environmental risk) → resilience–vulnerability equilibrium	Conceptual balancing perspective:
(i) clarifies the coexistence of resilience and vulnerability.
(ii) supports trade-off reasoning in resilience strategy formulation.
(iii) reinforces the need for prioritization of resilience drivers rather than unconditional maximization.
Shan et al. (2023)	E-commerce platforms	SEM	Behaviour–Capability– Performance (BCP) paradigm	Supply chain integration (supplier, internal, customer) → supply chain resilience (risk management, agility, robustness, restoration) → sustainable performance (economic, environmental, social)	Causal validation stream:
(i) empirically confirms integration → resilience → performance relationship.
(ii) legitimizes resilience as mediating capability.
(iii) supports importance of collaboration and information-sharing factors.
Russo et al. (2022)	Retail supply chains	Scenario-based experiments (ANCOVA; PROCESS macro)	Procedural Justice Theory (PJT) + Interactional justice	Service recovery resilience → Procedural justice → Customer satisfaction → Customer loyalty (moderated by interactional justice)	Customer-centric resilience perspective:
(i) demonstrates behavioural outcomes of recovery capability.
(ii) highlights communication and fairness as moderators of resilience outcomes.
(iii) extends resilience analysis to service recovery contexts.
Wang et al. (2023)	E-commerce supply chains	Empirical / case-based	Information Processing Theory + Organizational Ambidexterity	Information-processing fit → exploration/exploitation balance → agility, redundancy, flexibility capabilities	Capability development mechanism:
(i) explains how information-processing capability shapes resilience.
(ii) supports inclusion of agility, redundancy, and flexibility as key SCR capabilities.
(iii) provides theoretical grounding for interaction among resilience drivers.
Yan et al. (2023)	CBEC	Case study	Information Processing Theory	Information processing fit → 3 collaborative innovation modes (technology, management, market) →resilience capabilities (agility, redundancy)	Collaboration-driven resilience mechanism:
(i) supports the role of knowledge exchange and digital collaboration.
(ii) provides causal logic linking collaboration to resilience capabilities.
(iii) reinforces upstream drivers in resilience systems.
López et al. (2022)	Global fashion supply chain	Calibrated Fuzzy Best–Worst Method; expert judgement; sensitivity analysis	Decision-analytic resilience evaluation using fuzzy MCDM	Two-tier resilience evaluation framework: resilience capabilities and performance indicators; calibrated weighting scheme	Methodological benchmark:
(i) validates fuzzy expert-based weighting of resilience drivers.
(ii) supports methodological legitimacy of advanced fuzzy MCDM approaches.
(iii) provides comparative baseline for T-SF-based prioritization.
Banomyong et al. (2025)	Cross-border supply chains	Cross-Border Resilience Performance Index	Supply chain resilience capability perspective integrated with Logistics Performance Index concepts	Strategic capabilities: integrated flexibility & redundancy; integrated collaboration & agility → resilience performance index	Measurement benchmark:
(i) operationalizes resilience into measurable capability indicators.
(ii) supports factor-based resilience evaluation.
(iii) provides reference for quantitative resilience assessment frameworks.
Ma et al. (2025)	Cross-border logistics	System dynamics modelling (causal loops, stock–flow simulation)	Systems thinking and the feedback-loop resilience perspective	Dynamic interactions among inventory, transportation delay, capacity adjustment, demand fluctuation, and recovery rate	Dynamic interaction justification:
(i) demonstrates non-linear relationships among resilience drivers.
(ii) supports interaction-aware modelling.
(iii) reinforces the need for system-level resilience analysis beyond static factor evaluation.
Garola et al. (2023)	Express logistics during COVID-19	Multiple case studies	Resource-Based View and Dynamic Capabilities perspective	Resource reconfiguration under disruption; key enablers: visibility, communication, digital technologies, prior knowledge	Operational resilience evidence:
(i) validates micro-foundations of resilience in logistics networks.
(ii) highlights digitalization and communication as resilience enablers.
(iii) supports practical implications for logistics coordination and disruption response.

Note: “→” denotes a directional relationship (e.g., causal influence or hierarchical structure) between constructs.

2.2 Factors and Strategies Affecting CBECSC Resilience

This research used a multi-layered and systematic screening method to provide the final structure for this study—comprising three main dimensions, twelve sub-dimensions, and forty-nine factors Table 2. Factors affecting CBECSC resilience and Table 3. Strategies affecting CBECSC resilience are the results of a structured synthesis process based on established academic literature.

Table 2.
Factors Affecting CBECSC Resilience.

Main Dimensions Sub-Dimensions Factors References

D1-Supply Chain efficiency Supply chain agility F1 Trade Policies T. Chen et al. (2022)

F2 Awareness of major links Liu et al. (2021)

F3 Regulatory Compliance Safavi-Naini et al. (2004)

F4 Response time to disruption Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)

F5 Speed of recovery Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)

Crisis strategy F6 Reconfigure contingency strategy Yan et al. (2023), Wang et al. (2023), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)

F7 Forecast Accuracy López et al. (2022), Ayyildiz (2021)

F8 Disaster Planning Ayyildiz (2021)

F9 Crisis Planning Russo et al. (2022), Ayyildiz (2021)

F10 Recovery contingency strategy Wang et al. (2023), Shan et al. (2023), Zhang et al. (2023), Nabil et al. (2024)

F11 Reallocate assets strategy López et al. (2022), Shan et al. (2023), Zhang et al. (2023), Nabil et al. (2024)

Business improvement strategy F12 Strategic Fit Ayyildiz (2021)

F13 Strategic Stock Wang et al. (2023), Shan et al. (2023), Ayyildiz (2021)

F14 Evaluate key supplier risk Banomyong et al. (2025), Shan et al. (2023), Nabil et al. (2024)

F15 Detect risk signals Ma et al. (2025), Banomyong et al. (2025), Shan et al. (2023), Nabil et al. (2024)

F16 Assess key supplier impact Shan et al. (2023), Nabil et al. (2024)

Capacity for supply and demand F17 Designing generic product Yan et al. (2023), Nabil et al. (2024)

F18 Customize products Yan et al. (2023), Wang et al. (2023), Wang and Fang (2020)

F19 Utilize delay intelligently Wang et al. (2023), Nabil et al. (2024)

D2-Supply chain evolution Innovation capacity F20 Individual innovation potential Wang et al. (2023)

F21 Organizational innovation capability Yan et al. (2023), Nabil et al. (2024)

F22 Technology Level Garola et al. (2023), Yan et al. (2023), Ayyildiz (2021)

F23 Combined innovation ability Yan et al. (2023), Nabil et al. (2024)

Knowledge management F24 Gather social assets Wang et al. (2023), Liu et al. (2021)

F25 Utilize risk management experience Garola et al. (2023), Wang et al. (2023), Russo et al. (2022), Liu et al. (2021)

F26 Experience in risk response Garola et al. (2023), Wang et al. (2023), Russo et al. (2022), Nabil et al. (2024)

Collaboration among e-commerce SC partners F27 Knowledge exchange Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Zhang et al. (2023), Nabil et al. (2024)

F28 Interactive collaboration Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Zhang et al. (2023), Nabil et al. (2024)

F29 Digital collaboration Garola et al. (2023), Yan et al. (2023), Wang et al. (2023), Zhang et al. (2023), Ivanov (2020)

F30 Collaborative planning Banomyong et al. (2025), Wang et al. (2023), Zhang et al. (2023), Nabil et al. (2024)

Information sharing among partners F31 Positive business image Russo et al. (2022), Liu et al. (2021)

F32 Maintain information sharing agreements Banomyong et al. (2025), Wang et al. (2023), Russo et al. (2022), Zhang et al. (2023), Nabil et al. (2024)

F33 Integrity practices for real-time exchange Banomyong et al. (2025), Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Nabil et al. (2024)

D3-Supply Chain Adaptability Adaptive management F34 Long-term disruption capability Garola et al. (2023), Ma et al. (2025), Wang et al. (2023), Chowdhury and Quaddus (2017)

F35 Operation Agility Russo et al. (2022), Shan et al. (2023), Zhang et al. (2023), Ayyildiz (2021)

F36 Pre-disruption readiness Garola et al. (2023), Wang et al. (2023), Russo et al. (2022), Liu et al. (2021)

F37 Short-term disruption potential Ma et al. (2025), Russo et al. (2022), Liu et al. (2021)

Risk reduction practice F38 Mutual contingency plan Garola et al. (2023), Wang et al. (2023), Nabil et al. (2024)

F39 Collaborative risk team Garola et al. (2023), Wang et al. (2023), Kessler and Arlinghaus (2022)

F40 Risk Oversight Garola et al. (2023), Ma et al. (2025);Wang et al. (2023), Shan et al. (2023), Ayyildiz (2021)

F41 Collaborative environment Garola et al. (2023), Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Kessler and Arlinghaus (2022)

Supply chain flexibility F42 SC structure flexibility Banomyong et al. (2025), Wang et al. (2023), Zhang et al. (2023), Liu et al. (2021)

F43 Flexible Manufacturing System Ayyildiz (2021)

F44 Resource flexibility Banomyong et al. (2025), López et al. (2022), Wang et al. (2023), Ayyildiz (2021)

F45 Acquainted with SC management Russo et al. (2022), Nabil et al. (2024)

F46 Production Mix Flexibility Ma et al. (2025), Banomyong et al. (2025), Wang et al. (2023), Ayyildiz (2021)

Supply chain redundancy F47 Multiple vendor procurement Banomyong et al. (2025), Shan et al. (2023), Zhang et al. 2023), Kessler & Arlinghaus (2022)

F48 Safety stock inventory Banomyong et al. (2025), López et al. (2022), Wang et al. (2023), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)

F49 Multiple transport modes Banomyong et al. (2025), Yan et al. (2023), Wang et al. (2023), Kessler and Arlinghaus (2022)

Main Dimensions	Sub-Dimensions	Factors	References
D1-Supply Chain efficiency	Supply chain agility	F1 Trade Policies	T. Chen et al. (2022)
		F2 Awareness of major links	Liu et al. (2021)
		F3 Regulatory Compliance	Safavi-Naini et al. (2004)
		F4 Response time to disruption	Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)
		F5 Speed of recovery	Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)
	Crisis strategy	F6 Reconfigure contingency strategy	Yan et al. (2023), Wang et al. (2023), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)
		F7 Forecast Accuracy	López et al. (2022), Ayyildiz (2021)
		F8 Disaster Planning	Ayyildiz (2021)
		F9 Crisis Planning	Russo et al. (2022), Ayyildiz (2021)
		F10 Recovery contingency strategy	Wang et al. (2023), Shan et al. (2023), Zhang et al. (2023), Nabil et al. (2024)
		F11 Reallocate assets strategy	López et al. (2022), Shan et al. (2023), Zhang et al. (2023), Nabil et al. (2024)
	Business improvement strategy	F12 Strategic Fit	Ayyildiz (2021)
		F13 Strategic Stock	Wang et al. (2023), Shan et al. (2023), Ayyildiz (2021)
		F14 Evaluate key supplier risk	Banomyong et al. (2025), Shan et al. (2023), Nabil et al. (2024)
		F15 Detect risk signals	Ma et al. (2025), Banomyong et al. (2025), Shan et al. (2023), Nabil et al. (2024)
		F16 Assess key supplier impact	Shan et al. (2023), Nabil et al. (2024)
	Capacity for supply and demand	F17 Designing generic product	Yan et al. (2023), Nabil et al. (2024)
		F18 Customize products	Yan et al. (2023), Wang et al. (2023), Wang and Fang (2020)
		F19 Utilize delay intelligently	Wang et al. (2023), Nabil et al. (2024)
D2-Supply chain evolution	Innovation capacity	F20 Individual innovation potential	Wang et al. (2023)
		F21 Organizational innovation capability	Yan et al. (2023), Nabil et al. (2024)
		F22 Technology Level	Garola et al. (2023), Yan et al. (2023), Ayyildiz (2021)
		F23 Combined innovation ability	Yan et al. (2023), Nabil et al. (2024)
	Knowledge management	F24 Gather social assets	Wang et al. (2023), Liu et al. (2021)
		F25 Utilize risk management experience	Garola et al. (2023), Wang et al. (2023), Russo et al. (2022), Liu et al. (2021)
		F26 Experience in risk response	Garola et al. (2023), Wang et al. (2023), Russo et al. (2022), Nabil et al. (2024)
	Collaboration among e-commerce SC partners	F27 Knowledge exchange	Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Zhang et al. (2023), Nabil et al. (2024)
		F28 Interactive collaboration	Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Zhang et al. (2023), Nabil et al. (2024)
		F29 Digital collaboration	Garola et al. (2023), Yan et al. (2023), Wang et al. (2023), Zhang et al. (2023), Ivanov (2020)
		F30 Collaborative planning	Banomyong et al. (2025), Wang et al. (2023), Zhang et al. (2023), Nabil et al. (2024)
	Information sharing among partners	F31 Positive business image	Russo et al. (2022), Liu et al. (2021)
		F32 Maintain information sharing agreements	Banomyong et al. (2025), Wang et al. (2023), Russo et al. (2022), Zhang et al. (2023), Nabil et al. (2024)
		F33 Integrity practices for real-time exchange	Banomyong et al. (2025), Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Nabil et al. (2024)
D3-Supply Chain Adaptability	Adaptive management	F34 Long-term disruption capability	Garola et al. (2023), Ma et al. (2025), Wang et al. (2023), Chowdhury and Quaddus (2017)
		F35 Operation Agility	Russo et al. (2022), Shan et al. (2023), Zhang et al. (2023), Ayyildiz (2021)
		F36 Pre-disruption readiness	Garola et al. (2023), Wang et al. (2023), Russo et al. (2022), Liu et al. (2021)
		F37 Short-term disruption potential	Ma et al. (2025), Russo et al. (2022), Liu et al. (2021)
	Risk reduction practice	F38 Mutual contingency plan	Garola et al. (2023), Wang et al. (2023), Nabil et al. (2024)
		F39 Collaborative risk team	Garola et al. (2023), Wang et al. (2023), Kessler and Arlinghaus (2022)
		F40 Risk Oversight	Garola et al. (2023), Ma et al. (2025);Wang et al. (2023), Shan et al. (2023), Ayyildiz (2021)
		F41 Collaborative environment	Garola et al. (2023), Yan et al. (2023), Wang et al. (2023), Russo et al. (2022), Shan et al. (2023), Kessler and Arlinghaus (2022)
	Supply chain flexibility	F42 SC structure flexibility	Banomyong et al. (2025), Wang et al. (2023), Zhang et al. (2023), Liu et al. (2021)
		F43 Flexible Manufacturing System	Ayyildiz (2021)
		F44 Resource flexibility	Banomyong et al. (2025), López et al. (2022), Wang et al. (2023), Ayyildiz (2021)
		F45 Acquainted with SC management	Russo et al. (2022), Nabil et al. (2024)
		F46 Production Mix Flexibility	Ma et al. (2025), Banomyong et al. (2025), Wang et al. (2023), Ayyildiz (2021)
	Supply chain redundancy	F47 Multiple vendor procurement	Banomyong et al. (2025), Shan et al. (2023), Zhang et al. 2023), Kessler & Arlinghaus (2022)
		F48 Safety stock inventory	Banomyong et al. (2025), López et al. (2022), Wang et al. (2023), Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021)
		F49 Multiple transport modes	Banomyong et al. (2025), Yan et al. (2023), Wang et al. (2023), Kessler and Arlinghaus (2022)

Table 3.

Strategies Affecting CBECSC Resilience.

Strategies	Explanation	References
A1 Government Support	Preferential policies, tax incentives, human resource training, financial support, and infrastructure development can stabilize and improve the supply chain.	Zhang et al. (2023), Nakandala et al. (2025)
A2 Speeding Up Recovery	Using real-time monitoring, emergency response training, and better decision-making to safeguard supply networks. Recovering quickly lowers disruption.	Garola et al. (2023), Liu et al. (2021), Nabil et al. (2024)
A3 Implementing Contingency Plans and Backup Plans	Safety stockpiles, contingency planning, and alternate logistics are crucial. Contingency preparations and backup systems help reduce disruptions, delays, and financial losses, thereby strengthening SCRE.	Banomyong et al.(2025), Liu et al. (2021), Nabil et al. (2024)
A4 Promoting Collaboration and Fairness	The fairness and collaboration of supply chain partners build trust and transparency by fostering open communication and promoting fair agreements.	Shan et al. (2023), Zhang et al. (2023), Liu et al. (2021), Nabil et al. (2024)
A5 Strengthening international coordination	International alliances, free trade agreements, and process standardization boost cross-border supply chain operations.	Banomyong et al. (2025), Liu et al. (2021), Nabil et al. (2024)
A6 Building risk management team	Creating a risk management department, training employees, and evaluating past data can help a company respond.	Wang et al. (2023), Liu et al. (2021), Nabil et al. (2024)

D1-Supply chain efficiency emphasizes the speed and responsiveness of the supply network. Four categories of supply chain efficiency include agility, crisis strategy, business improvement plan, and supply and demand capacity (Liu et al., 2021; Nabil et al., 2024).

D2-Supply chain evolution is underscored by the modernization and enhancement of supply chain recovery and reconfiguration speed. Experts divide this dimension into four sub-factors: innovation capability, knowledge management, e-commerce SC partner collaboration, and partner information exchange (Liu et al., 2021; Nabil et al., 2024).

D3-Supply chain adaptability refers to a supply chain's ability to undergo various transitions and recover efficiently after disruptions. This dimension encompasses adaptive management, risk reduction strategies, supply chain flexibility, and redundancy (Liu et al., 2021; Nabil et al., 2024).

2.3 Overview of MCDM Techniques and Applications in SCRE

Multidimensional statistics, linear functions, software, and economic theories have become part of today's decision-making toolkit. In reality, choices can be rational or irrational, often resting on assumptions and shaped by factors such as biology, culture, society, and physiology. Among these approaches, MCDM stands out as a breakthrough, highly accurate method that has been in use since the 1950s to test mathematical models. Its theoretical framework helps organize problems and capture decision-makers’ preferences (Lu et al., 2018). It is beneficial for complex, multi-factor problems where the best solution must align with stated priorities, often through expert panels ranking each criterion. In the case of CBECSC resilience, entrepreneurs, the government, and institutions all play a role (Nabil et al., 2024). For this reason, our study applies MCDM to examine resilience across several domains. This study requires precise and comprehensive methods, such as MCDM, to examine CBECSC resilience across multiple domains. For example, Liu et al. (2023) used an integrated MCDM combining ISM, Fuzzy MICMAC, and DEMATEL to identify, classify, and analyze hierarchical and causal relationships between factors driving customer-oriented maritime supply chain resilience (MSCR) during the COVID-19 pandemic. Similarly, Das et al. (2022) used MCDM to identify relevant SCRE factors during the COVID-19 pandemic. AHP sorts things by the extent to which they affect the outcome, while DEMATEL identifies the major causes by examining how they interact with one another.

2.4 Overview of Fuzzy-MCDM Techniques and Applications in SCRE

MCDM supports engineers in making decisions in a structured, transparent, and methodical manner. However, independent decision-makers often face uncertainty, ambiguity, and incorrect information. Fuzzy Multi-Criteria Decision-Making (FMCDM), a blend of MCDM and Fuzzy Set Theory, tackles these issues. According to previous research, fuzzy numbers can effectively represent complex and ambiguous information (Jafarzadeh Ghoushchi et al., 2019). Since Zadeh (1965) introduced Fuzzy Sets, several academics have explored various applications to manage multi-criteria evaluations. Atanassov (1986) created the Intuitionistic Fuzzy Sets (IFS) in 1986. Then Yager (2017) developed the q-Rung arctograph set in 2013 and Pythagorean Fuzzy Sets (PyFS) in 2017. Cuong (2014) structured the Picture Fuzzy Sets (PFS). Recently, Gündoğdu (2019) introduced SFS, and Mahmood et al. (2019) further developed the SFS into T-SFS. Since the launch of Fuzzy Sets, their integration with MCDM methods has mitigated the deficiencies in information processing that traditional methods cause. Ahmed Mohammed et al. (2019) designed a sustainable and resilient supply chain network in the UK by combining Fuzzy multi-objective optimization. They applied Fuzzy AHP to weight the resilience pillars, optimized infrastructure for cost, environmental impact, and resilience, and used TOPSIS to select the best Pareto solution. Liu et al. (2023) examined MSCR during the COVID-19 pandemic by utilizing MCDM, employing ISM to establish factor hierarchies, Fuzzy MICMAC to evaluate influence-dependence, and DEMATEL to differentiate between causes and effects in China. Haldar et al. (2012) utilized the AHP–TOPSIS–QFD multi-method analysis framework to develop a quantitative approach to the supplier selection problem in the context of disruptions and natural disasters. Rehman et al. (Rehman & Ali, 2022) prioritized healthcare SCRE techniques by using Fuzzy AHP to weight risks, Fuzzy TOPSIS to rank them, and Fuzzy QFD to find the best solutions. López et al. (2022) enhanced resilience for a European shoe firm in Latin America by using the Calibrated Fuzzy Best–Worst Method (CFBWM), which combines the CBWM approach to refine fuzzy values and the FBWM method to reduce evaluation ambiguity. Overall, many resilience studies integrate fuzzy and MCDM techniques to address multi-factor challenges.

Although SCR is a topic of considerable research, none of the studies utilize T-SF, one of the most recent Fuzzy Sets developed based on SFS. Although SFS can manage information more effectively than traditional fuzzy methods throughout the sphere, some information fields, upon examination, exceed the coverage capacity of the sphere; thus, T-SF is utilized as an efficient technique to address the limitations of SFS. T-SF has three membership functions $(α, \; \; β, \; γ)$ with the sum of the power t of the membership function in the range of [0, 1] $(0 \leq α^{t} + β^{t} + γ^{t} \leq 1)$ with $t \; \in N^{*}$ ( $N^{*}$ is the set of natural numbers excluding zero. According to Mahmood et al. (2019), there are some T-SF's characteristics:

When $t = 1 \to PFS$ .

When $t = 2 \to SFS$ .

When $β = 0 \to q - ROPFS$ .

When $t = 2$ and $β = 0 \to PyFS$ .

When $t = 1$ and $β = 0 \to IFS$ .

When $t = 1$ , $β = 0$ and $γ = 0 \to Traditional Fuzzy$ .

Similar to SFS, T-SFS represents expert evaluations using three components: membership, abstinence, and non-membership, while the refusal degree is derived from the relationship among these components. The key advancement of T-SFS lies in the introduction of the exponent parameter $(t)$ , which enlarges the feasible evaluation space compared with SFS and enables decision makers to express uncertainty and hesitation more flexibly. In complex information-processing methods such as DEMATEL, the aggregation and propagation of influence values may produce fuzzy evaluations that exceed the quadratic constraint of SFS. In such situations, T-SFS can accommodate these evaluations by adjusting the parameter $(t)$ , avoiding normalization procedures required in earlier fuzzy frameworks. This flexibility is particularly valuable in CBEC logistics, where expert judgments often involve incomplete information, rapidly changing trade policies, and uncertain disruption risks.

Due to its versatility, T-SFS is widely used in uncertain data processing and MCDM. Many researchers combine T-SF with MCDM to solve problems. Fetanat and Tayebi (2024) employed the T-shaped fuzzy set-based ordinal priority approach (T-SF-OPA) to develop a support system for robustly determining the most suitable thermal solar technology for integration into enhanced oil recovery (EOR) processes, based on fundamental principles of sustainability and availability under uncertainty. Özdemirci et al. (2023) emphasize the importance of risk analysis and qualified personnel, utilizing the T-SF TOP-DEMATEL method to determine the significance of factors influencing interest-free social banking systems and employing causality analysis to assess their impact. With T-SF MCDM, Aytekin et al. (2024) identified the optimal sustainable strategy and addressed the challenges of green energy in logistics companies. Intercriteria correlation integrates subjective and objective weights in SOWIA. The T-SF-ARAS approach evaluates the additive ratio system, and the T-SF CRITIC method measures the importance of criteria. In their study on criteria in cooperative intelligent transportation systems (C-ITS), Anjum et al. (2024) employed the T-SF technique to enhance precision and manage uncertainty, in conjunction with the CRITIC and Weighted Aggregated Sum Product Assessment (WASPAS) method. WASPAS created weighted scores for ideal selections, thereby enhancing the sustainability of the transportation system, whereas CRITIC examined the correlations between the criteria.

2.5 Overview of Fuzzy-MCDM Techniques and Applications in CBECSC Resilience

CBECSC resilience has emerged as a significant concern in global commerce, particularly in the context of crises such as the Russia-Ukraine conflict and the COVID-19 pandemic. The pandemic disrupted travel, distribution, and production worldwide, while the slowdown in China—the world's largest industrial hub—further strained supply chains, causing shortages, rising shipping costs, and inconsistent raw material supplies (Das et al., 2022). The Russia-Ukraine conflict has affected wheat, sunflower oil, corn, natural gas, oil, and fertilizer. Sanctions and trade restrictions have tripled or quadrupled food shortages and fuel costs, particularly in countries that export agricultural and petroleum products from the region, leading to economic instability and global concerns about food security (Zhou et al., 2023). For example, Qi et al. (2023) assessed SCRE using data from a Chinese e-commerce platform, highlighting operational indicators, information exchange, strategic planning, and logistical collaboration. CBECSC resiliency studies have successfully employed MCDM approaches. Liu et al. (2021) employed Fuzzy DEMATEL-ISM to assess the comprehensive influence degree, causal link, and logical hierarchy of all CBSCR-affecting factors. Information Processing Theory, High Reliability Theory, and Social Exchange Theory proposed 36 CBSCR-affecting elements in their ternary theory of SCR research. Zhang et al. (2023) investigated how CBEC firms achieve an ideal equilibrium between SCRE and vulnerability using the Fuzzy AHP and Fuzzy TOPSIS methodologies, along with the EWM, SAW, and ISM. Nabil et al. (2024) developed a hybrid model to examine the resilience of transnational e-commerce supply chains (TNSCRE). The EWM establishes criteria weights, whereas the SAW ranks components based on weighted scores. The ISM assesses fundamental component interconnections. Vietnam's CBEC companies are primarily small and medium-sized enterprises with logistical systems still in development, which have suffered from global swings due to limited financial resources, skilled labor shortages, and a lack of digital understanding. Thus, CBECSC resilience research in Vietnam is crucial given the changing scenario.

Based on the above discussion, it can be observed that MCDM techniques have been widely and extensively applied in studies on SCRE in general, and CBECSC resilience in particular. The selection of an appropriate set of MCDM tools for the research framework should therefore be carefully determined based on the advantages and limitations of each method. In this regard, Table 4 presents a comparative overview of prominent MCDM techniques to justify their suitability for the present study.

Table 4.
The Comparison of MCDM Methods.

Method Advantages Limitations References Suitability

Weighted Sum Method (WSM) Straightforward and easy to use; simply combines multiple goal functions into a single objective with a weight vector. Unable to manage different criteria with different priority levels. Cohen et al. (2022) Not suitable

Criteria Importance Through Intercriteria Correlation (CRITIC) Objective and purely statistical; determines criteria weights from intercriteria correlation without involving experts. Disregards the objectives and preferences of decision-makers. Tuş and Aytaç Adalı (2019) Not suitable

Evaluating based on Distance from Average Solution (EDAS) Practical for situations where criteria conflict; Evaluation based on the distance from the mean of the solution. Calculations are complex and lengthy Xuan et al. (2022) Limited

Best Worst Method (BWM) Needs fewer pairwise comparisons, resulting in more consistent outcomes; uses a nonlinear min–max model to assign weights. Involves nonlinear optimization and requires a lot of computational capacity. Rezaei et al. (2015) Limited

AHP Has a clear hierarchical structure; easier to follow than some other methods; well-established in practice. The interconnectedness of criteria cannot be captured. Kaya et al. (2018) Not suitable

Delphi Gathers expert judgment effectively; an anonymous and structured evaluation process helps synthesize opinions, eliminating the influence of individual collaborators, conflicting viewpoints, and pressure groups; no restriction on the number of variables considered. Does not measure the influence between factors. Dohale et al. (2021) Suitable (for screening)

ISM Reveals the structural links between components. Lacks the ability to quantify influence strength. Beikkhakhian et al. (2015) Limited

DEMATEL Identifies cause–effect relationships and measures their strength; pinpoints key criteria; separates causes from effects; better than ISM for quantifying influence; allows criteria to be mapped hierarchically. Depends on professional judgment and is receptive to personal opinions. Er-retby et al. (2025) Highly suitable

COCOSO Thorough, transparent, and adaptable; can balance multiple criteria and adjust weights to suit preferences; reconciles conflicting objectives for optimal performance; builds trust through decision-making transparency. Requires normalized data; relatively new method. Ortíz-Barrios et al. (2025) Highly suitable

Method	Advantages	Limitations	References	Suitability
Weighted Sum Method (WSM)	Straightforward and easy to use; simply combines multiple goal functions into a single objective with a weight vector.	Unable to manage different criteria with different priority levels.	Cohen et al. (2022)	Not suitable
Criteria Importance Through Intercriteria Correlation (CRITIC)	Objective and purely statistical; determines criteria weights from intercriteria correlation without involving experts.	Disregards the objectives and preferences of decision-makers.	Tuş and Aytaç Adalı (2019)	Not suitable
Evaluating based on Distance from Average Solution (EDAS)	Practical for situations where criteria conflict; Evaluation based on the distance from the mean of the solution.	Calculations are complex and lengthy	Xuan et al. (2022)	Limited
Best Worst Method (BWM)	Needs fewer pairwise comparisons, resulting in more consistent outcomes; uses a nonlinear min–max model to assign weights.	Involves nonlinear optimization and requires a lot of computational capacity.	Rezaei et al. (2015)	Limited
AHP	Has a clear hierarchical structure; easier to follow than some other methods; well-established in practice.	The interconnectedness of criteria cannot be captured.	Kaya et al. (2018)	Not suitable
Delphi	Gathers expert judgment effectively; an anonymous and structured evaluation process helps synthesize opinions, eliminating the influence of individual collaborators, conflicting viewpoints, and pressure groups; no restriction on the number of variables considered.	Does not measure the influence between factors.	Dohale et al. (2021)	Suitable (for screening)
ISM	Reveals the structural links between components.	Lacks the ability to quantify influence strength.	Beikkhakhian et al. (2015)	Limited
DEMATEL	Identifies cause–effect relationships and measures their strength; pinpoints key criteria; separates causes from effects; better than ISM for quantifying influence; allows criteria to be mapped hierarchically.	Depends on professional judgment and is receptive to personal opinions.	Er-retby et al. (2025)	Highly suitable
COCOSO	Thorough, transparent, and adaptable; can balance multiple criteria and adjust weights to suit preferences; reconciles conflicting objectives for optimal performance; builds trust through decision-making transparency.	Requires normalized data; relatively new method.	Ortíz-Barrios et al. (2025)	Highly suitable

Among various MCDM approaches, the Delphi method is widely recognized as an effective technique for screening and validating factors based on expert consensus. This method enables the systematic collection of expert opinions through a structured and anonymous evaluation process, which facilitates the aggregation of diverse perspectives while minimizing the influence of dominant individuals, conflicting viewpoints, or group pressure. In addition, Delphi does not impose restrictions on the number of variables considered, making it particularly suitable for exploratory studies involving complex systems such as supply chains. As a result, the Delphi method is frequently employed as a preliminary step to identify and refine a set of critical factors before conducting more advanced analytical procedures. However, a key limitation of the Delphi method is its inability to reveal the interrelationships among the identified factors. In complex systems such as supply chains, the influencing factors rarely operate independently; instead, they interact and form intricate causal structures. To address this limitation, the DEMATEL method is employed to analyze the causal relationships among factors and to quantify the strength of their mutual influences. DEMATEL enables the identification of key driving factors, distinguishes between cause and effect groups, and provides a visual representation of the structural relationships within the system. Compared with other structural methods such as ISM, DEMATEL offers stronger capabilities for quantifying influence intensity and revealing complex cause–effect relationships among criteria. Consequently, the integration of Delphi and DEMATEL forms a complementary analytical framework in which Delphi is used to identify and validate relevant factors, while DEMATEL is employed to analyze their causal interactions and structural relationships (Gupta & Gupta, 2026). In the context of this study, CBECSC resilience represents a multi-criteria decision problem involving multiple dimensions, including logistics infrastructure, trade policies, technological capabilities, and risk management strategies, as well as multiple stakeholders. These factors are not only numerous but also highly interdependent. Therefore, the combined application of Delphi and DEMATEL provides a suitable framework for selecting, evaluating, and analyzing the critical factors influencing CBECSC resilience in Vietnam.

Furthermore, the COCOSO method is adopted to prioritize resilience-enhancing strategies. Evaluating CBECSC resilience involves multiple criteria with varying levels of influence, and tradeoffs often exist among these criteria. The COCOSO method is based on the concept of a compromise solution, integrating two aggregation logics: the SAW model and the Weighted Product Model (WPM). This hybrid mechanism allows the model to balance different evaluation perspectives and identify an optimal compromise solution within complex decision-making environments (Q. Y. Chen et al., 2022). Such characteristics make COCOSO particularly suitable for addressing complex decision problems such as assessing the resilience of CBECSC. Another important advantage of COCOSO lies in the stability of its ranking results, as the method is less sensitive to the addition or removal of alternatives compared with several traditional MCDM techniques (Cui et al., 2021). This property enhances the robustness and reliability of decision-making outcomes. This feature is especially relevant in the CBECSC context, where the set of factors or strategies may evolve over time due to changes in market conditions, trade policies, or logistics infrastructure development.

However, traditional MCDM methods have often been criticized for their limited ability to handle vague and uncertain information in real-world decision environments (López et al., 2022). These approaches typically rely on precise numerical inputs, which may not adequately capture the ambiguity and hesitation inherent in expert judgments. This limitation is particularly evident in complex systems such as supply chains, where many factors are interdependent and evaluations are often qualitative. To address this issue, many studies have integrated MCDM techniques with fuzzy set theories to better represent uncertainty in expert assessments (Zhang et al., 2023). In this study, the Delphi, DEMATEL, and COCOSO methods are combined with the T-SF. framework. As an advanced fuzzy representation, T-SF generalizes several earlier fuzzy models, including Fuzzy Sets, IFS, and PyFs, allowing a more comprehensive representation of uncertainty. This integration enhances the robustness and reliability of resilience evaluation in CBECSC.

2.6 Research Gaps

Although CBECSC resilience research is growing, several gaps remain. These gaps require new approaches and regional studies, especially in Vietnam's e-commerce market. An extensive literature review identified the following research needs: Vietnam lacks research on CBECSC resilience, despite worldwide interest. Previous research has focused on mature e-commerce economies, such as China (Qi et al., 2023; Yan et al., 2023), or taken a transnational perspective (Nabil et al., 2024). Vietnam's growing logistics infrastructure, regulatory frameworks, and cross-border trade rules necessitate targeted investigations. CBECSC in Vietnam cannot create targeted disruption mitigation and resilience measures without region-specific research. Most assessments of resilience factors rely on expert opinions without differentiating input weights based on subject matter expertise or experience (Yan et al., 2023). Standardizing expert contributions diminishes the accuracy and reliability of MCDM analysis. This oversight is crucial in complex areas such as CBECSC resilience, where specialized insights significantly impact factor and strategy prioritization.

Fuzzy-enhanced MCDM is often utilized in resilience studies, although T-SF has not been applied to CBECSC resilience. The latest fuzzy logic innovation, T-SFS, adds reluctance for more complex uncertainty representation. Classic Fuzzy Sets and their earlier expansions, such as IFS and SFS (Govindan et al., 2015; Kutlu Gündoğdu & Kahraman, 2019). Struggle to reflect the dynamic and uncertain environments of CBECSC. Many researchers use unvalidated single-method or hybrid models. Liu et al. (2021) used DEMATEL-ISM without cross-validation with complementary methods, and Liu et al. (2021) utilized an integrated MCDM approach combining ISM, Fuzzy MICMAC, and DEMATEL to identify, classify, and analyze hierarchical and causal relationships between factors, without considering alternatives. This lack of scientific rigor limits the reliability and generalizability of CBECSC resilience research.

This is the first T-SFS and MCDM investigation of CBECSC resilience variables in Vietnam. This survey also weighted experts by research topic competency. This research fills these gaps to build CBECSC resilience theory and techniques, providing practitioners and policymakers with actionable insights to enhance Vietnam's global standing in CBEC.

3 Methodology

3.1 Overview of Fuzzy Set Theories

3.1.1 Ordinary Fuzzy Sets (FSs)

Zadeh (1965) presented Fuzzy sets as entities characterized by a spectrum of membership grades. He defined concepts such as inclusion, union, intersection, complement, relation, convexity, and other properties of Fuzzy sets. Zadeh (1972) introduced linguistic operators on a vaguely defined set that represents the meaning of its operands.

A fuzzy set in X is a set of ordered pairs if X is a collection of items generally indicated by $x$ :

\tilde{A} = {(x, α_{\tilde{A}} (x) | x \in X)},

(1)

where

α_{\tilde{A}} (x)

is referred to as the membership function. The set of non-negative real numbers with a finite supremum defines its range.

3.1.2 Intuitionistic Fuzzy Sets (IFSs)

The membership degree (MD) and non-membership degree (NMD) values for every component x in X are included in are incorporated in Atanassov's IFS ensuring that the sum of MD and NMD does not exceed 1 (Atanassov, 1986). $X \neq \emptyset$ be a given set. An IFS in X is an object $\tilde{A}$ given by:

\tilde{A} = {⟨ x, α_{\tilde{A}} (x), β_{\tilde{A}} (x) ⟩; x \in X},

(2)

where

α_{\tilde{A}} : X \to [0, 1]

and

β_{\tilde{A}} : X \to [0, 1]

and

0 \leq α_{\tilde{A}} (x) + β_{\tilde{A}} (x) \leq 1 for every x \in X .

(3)

The IF index or hesitation degree of whether or not $x$ belongs to $S$ is $γ_{S} (x)$ , the third parameter of the IFS.

\begin{aligned} γ_{\tilde{A}} & = 1 - [α_{\tilde{A}} (x) + β_{\tilde{A}} (x)], for every x \in X, \end{aligned}

(4)

\begin{aligned} 0 & \leq γ_{\tilde{A}} (x) \leq 1. \end{aligned}

(5)

3.1.3 Pythagorean Fuzzy Sets (PyFSs)

The notion of PyFSs, proposed by Yager (2014) is delineated as follows. Let X be a space of points (objects) and $x \in X$ . A PyFS A in X is defined by a MD $α_{A} (x)$ , and NMD $β_{A} (x$ ).

A = {⟨ x, α_{A} (x), β_{A} (x) ⟩; x \in X},

(6)

where

α_{A} (x) \to [0, 1]

, and

β_{A} (x) \to [0, 1]

, and

0 \leq α_{A}^{2} (x) + β_{A}^{2} (x) \leq 1, for x \in X .

(7)

The hesitation degree $γ_{A} (x)$ of x to A is defined as:

γ_{A} (x) = \sqrt{1 - (α_{A}^{2} (x) + β_{A}^{2} (x))} .

(8)

3.1.4 q-Rung Orthopair Fuzzy Sets (qROFs)

The extensions of PyFSs referred to as q-Rung Orthopair Fuzzy Sets (qROFs) are elucidated through the foundational concepts of q-ROFs.

Assuming X is a reference set, then a q-ROF can be described as:

A = {⟨ x, α_{A} (x), β_{A} (x) ⟩; x \in X},

(9)

where MD

α_{A} (x) \to [0, 1]

, and NMD

β_{A} (x) \to [0, 1]

with

0 \leq α_{A}^{q} (x) + β_{A}^{q} (x) \leq 1, q \geq 1

(10)

The hesitation degree $γ_{A} (x)$ can be computed as follows using the MD and NMD:

γ_{A} (x = \sqrt[q]{α_{A}^{q} (x) + β_{A}^{q} (x) - (α_{A}^{q} (x) \cdot β_{A}^{q} (x))} .

(11)

3.1.5 Picture Fuzzy Sets (PFSs)

Atanassov (1986) expanded Fuzzy sets by including the degree of non-membership; yet, evaluations reliant on IFS can occasionally be nonsensical. PFS theory (Cuong, 2014), which uses Fuzzy Sets and Interval-valued Fuzzy Sets to articulate decision makers’ affirmative, abstention, negative, and refusal signals.

Let PFS A on a universe X be an object in the form of

A = {⟨ x, α_{A} (x), β_{A} (x), γ_{A} (x) ⟩; x \in X},

(12)

where

α_{A} (x)

β_{A} (x)

, and

γ_{A} (x)

denotes the degree of positive membership; neutral membership and negative membership of x in A, respective, and

α_{A} (x) \in [0, 1], β_{A} (x) \in [0, 1]

, and

γ_{A} (x) \; \in [0, 1]

α_{A} (x), β_{A} (x)

, and

γ_{A} (x)

satisfy the following conditions:

0 \leq α_{A} (x) + β_{A} (x) + γ_{A} (x) \leq 1.

(13)

3.1.6 Spherical Fuzzy Sets (SFSs)

SFSs are an extension of Fuzzy Sets developed by Kutlu Gündoğdu and Kahraman (2020), wherein each spherical fuzzy number denotes the membership degree (MD) $(α)$ , abstention degree (AD) $(β)$ , and non-membership degree (NMD) $(γ)$ inside the interval [0,1]. An SFS ${\tilde{A}}_{s}$ can be expressed as

{\tilde{A}}_{s} = {x, (α_{{\tilde{A}}_{s}} (x), β_{{\tilde{A}}_{s}} (x), γ_{{\tilde{A}}_{s}} (x)) | x \in X},

(14)

where

α_{{\tilde{A}}_{s}} (x) : X \to [0, 1], β_{{\tilde{A}}_{s}} (x) : X \to [0, 1], γ_{{\tilde{A}}_{s}} (x) : X \to [0, 1]

and

α_{A}^{2} (x) + β_{A}^{2} (x) + γ_{A}^{2} (x) \leq 1

with

\forall x \in X

3.2 T-Spherical Fuzzy

The following list includes key T-SFS terms and concepts by Mahmood et al. (2019) for the planned study.

Definition 1:
Let X be a nonempty fixed set; then a T-SFS T on X can be articulated as equation (15):
$T = {⟨ x, (m_{T} (x), n_{T} (x), p_{T} (x)) ⟩ | x \in X}$
(15)
Let $m_{T} (x) : X \to [0, 1], n_{T} (x) : X \to [0, 1]$ and $p_{T} (x) : X \to [0, 1]$ represent the degree of membership, degree of abstinence and degree of non-membership of x in T, respectively, which fulfill the following criterion for every $x \in X : 0 \leq m_{T}^{t} (x) + n_{T}^{t} (x) + p_{T}^{t} (x) \leq 1, t \geq 1$ .

The triplet $⟨ m, n, p ⟩$ is called T-spherical fuzzy number (T-SFN). Based on Mahmood et al. (2019), there are some T-SF's characteristics:

When $t = 1,$ it becomes PFS.

When $t = 2,$ it becomes SFS.

When $n = 0$ , it becomes q-ROPFS.

When $t = 2$ and $n = 0$ it becomes PyFS.

When $t = 1$ and $n = 0,$ $it becomes$ IFS.

When $t = 1,$ $n = 0,$ and $p = 0$ , $it becomes$ Traditional Fuzzy.
Definition 2:
Let $T^{'} = ⟨ m, n, p ⟩$ be a T-SFN, then the score function is given as equation (16):
$S (T^{'}) = (\frac{1 + m^{t} + n^{t} + p^{t}}{2}) .$
(16)
Clearly, $S (T^{'})$ is inside the interval $[0, 1]$ , and an increase in the score value of $S (T^{'})$ correlates with a higher T-SFN for $T^{'}$ .
Definition 3:
The accuracy function $H (T^{'})$ of a T-SFN is defined as equation (17):
$H (T^{'}) = m^{t} + n^{t} + p^{t} .$
(17)
The higher the precision value of $H (T^{'})$ , the greater the T-SF number $T^{'}$ will be.
Definition 4:
The comparison of $T_{1} = (m_{1}, n_{1}, p_{1})$ and $T_{2} = (m_{2}, n_{2}, p_{2})$ is as follows:
If $S (T_{1}) < S (T_{2})$ , then $T_{1} < T_{2}$ ,

If $S (T_{1}) > S (T_{2})$ , then $T_{1} > T_{2}$ ,

If $S (T_{1}) = S (T_{2})$ , and
If $H (T_{1}) < H (T_{2})$ , then $T_{1} < T_{2}$ ,

If $H (T_{1}) > H (T_{2})$ , then $T_{1} > T_{2}$ ,

If $H (T_{1}) = H (T_{2})$ , then $T_{1} = T_{2}$ .

Definition 5:
The operations of T-SFNs (López et al., 2022) from equations (18) to (21), Let $T_{1} = (m_{1}, n_{1}, p_{1})$ and $T_{2} = (m_{2}, n_{2}, p_{2})$

Addition $T_{1} \oplus T_{2}$
$= (\begin{array}{l} \sqrt[t]{1 - (1 - m_{1}^{t}) (1 - m_{2}^{t})}, \sqrt[t]{(1 - m_{1}^{t}) (1 - m_{2}^{t}) - (1 - m_{1}^{t} - n_{1}^{t}) (1 - m_{2}^{t} - n_{2}^{t})}, \\ \sqrt[t]{(1 - m_{1}^{t} - n_{1}^{t}) (1 - m_{2}^{t} - n_{2}^{t}) - (1 - m_{1}^{t} - n_{1}^{t} - p_{1}^{t}) (1 - m_{2}^{t} - n_{2}^{t} - p_{2}^{t})} \end{array}) .$
(18)

Multiplication $T_{1} \otimes T_{2}$
$= (\begin{array}{l} \sqrt[t]{(1 - p_{1}^{t} - n_{1}^{t}) (1 - p_{2}^{t} - n_{2}^{t}) - (1 - p_{1}^{t} - n_{1}^{t} - m_{1}^{t}) (1 - p_{2}^{t} - n_{2}^{t} - m_{2}^{t})}, \\ \sqrt[t]{(1 - p_{1}^{t}) (1 - p_{2}^{t}) - (1 - p_{1}^{t} - n_{1}^{t}) (1 - p_{2}^{t} - n_{2}^{t})}, \sqrt[t]{1 - (1 - p_{1}^{t}) (1 - p_{2}^{t})} \end{array}) .$
(19)

Multiplication by α scalar $(λ > 0)$ ; λ $T_{1}$
$= (\sqrt[t]{(1 - {(1 - m_{1}^{t})}^{λ})}, \sqrt[t]{{(1 - m_{1}^{t})}^{λ} -, {(1 - m_{1}^{t} - n_{1}^{t})}^{λ}}, \sqrt[t]{{(1 - m_{1}^{t} - n_{1}^{t})}^{λ} - {(1 - m_{1}^{t} - n_{1}^{t} - p_{1}^{t})}^{λ}}) .$
(20)

Power $(λ > 0)$ ; $T_{1}^{λ}$
$= (\sqrt[t]{{(1 - p_{1}^{t} - n_{1}^{t})}^{λ} - {(1 - p_{1}^{t} - n_{1}^{t} - m_{1}^{t})}^{λ}}, \sqrt[t]{{(1 - p_{1}^{t})}^{λ} - {(1 - p_{1}^{t} - n_{1}^{t})}^{λ}}, \sqrt[t]{(1 - {(1 - p_{1}^{t})}^{λ})}) .$
(21)

Definition 6:
The Minkowski distance between $T_{1} and T_{2}$ is defined as equation (22):
$d_{k} (T_{1}, T_{2}) = ({| m_{1}^{t} - m_{2}^{t} |}^{k} + {| n_{1}^{t} - n_{2}^{t} |}^{k} + {| p_{1}^{t} - p_{2}^{t} |}^{k})^{\frac{1}{k}},$
(22)
Where:
$t = 1$ represents the Hamming distance ;

$t = 2$ represents the clidean distance;

$t = + \infty$ represents the Chebychev distance .

Definition 7:
Let $T_{j} = ⟨ m_{j}, n_{j}, p_{j} ⟩ (j = 1, 2, \dots, n)$ be a collection of T-SFNs, $w = (w_{1}, w_{2}, \dots, w_{n})^{T}$ be the weight vector of $T_{j} (j = 1, 2, \dots, n)$ , such that $w_{j} \in [0, 1]$ and $\sum_{j}^{n} w_{j} = 1$ . Then, the T-spherical fuzzy weighted averaging interaction (T-SFWAI) operator is given as equation (23):
$\begin{array}{l} T - SFWAI (T_{1}, T_{2}, \dots, T_{n}) \\ = (\sqrt[t]{(1 - \prod_{j = 1}^{n} {(1 - m_{j}^{t})}^{w_{j}})}, \sqrt[t]{(\prod_{j = 1}^{n} {(1 - m_{j}^{t})}^{w_{j}} - \prod_{j = 1}^{n} {(1 - m_{j}^{t} - n_{j}^{t})}^{w_{j}}),} \sqrt[t]{(\prod_{j = 1}^{n} {(1 - m_{j}^{t} - n_{j}^{t})}^{w_{j}} - \prod_{j = 1}^{n} {(1 - m_{j}^{t} - n_{j}^{t} - p_{j}^{t})}^{w_{j}})}) . \end{array}$
(23)
Definition 8:
The T-spherical fuzzy weighted geometric interaction (T-SFWGI) operator is given as equation (24):
$\begin{array}{l} T - SFWGI (T_{1}, T_{2}, \dots, T_{n}) \\ = (\sqrt[t]{(\prod_{j = 1}^{n} {(1 - p_{j}^{t} - n_{j}^{t})}^{w_{j}} - \prod_{j = 1}^{n} {(1 - p_{j}^{t} - n_{j}^{t} - m_{j}^{t})}^{w_{j}})}, \sqrt[t]{(\prod_{j = 1}^{n} {(1 - p_{j}^{t})}^{w_{j}} - {(1 - p_{j}^{t} - n_{j}^{t})}^{w_{j}})}, (1 - \prod_{j = 1}^{n} \sqrt[t]{{(1 - p_{j}^{t})}^{w_{j}}})) . \end{array}$
(24)
3.3 T-Spherical Fuzzy Delphi (T-SF Delphi)

Step 1: Identify the primary factors.

In the first step, an extensive literature review is conducted to identify potential factors influencing the resilience of CBECSC. These factors, summarized in Table 2, serve as the basis for designing the expert evaluation questionnaire. Experts are then invited to assess the importance of each factor using the T-SF linguistic scale presented in Table 5.

Table 5.
The T-SF Linguistic Scale.

Scale $(m, n, p)$ Score Index (SI)

Absolutely more significance (AMS) (0.9, 0.1, 0.1) 9

Very high significance (VS) (0.8, 0.2, 0.2) 7

High significance (HS) (0.7, 0.3, 0.3) 5

Slightly more significance (SMS) (0.6, 0.4, 0.4) 3

Equally significance (ES) (0.5, 0.5, 0.5) 1

Slightly significance (SS) (0.4, 0.6, 0.4) 1/3

Low significance (LS) (0.3, 0.7, 0.3) 1/5

Very low significance (VLS) (0.2, 0.8, 0.2) 1/7

Absolutely low significance (ALS) (0.1, 0.9, 0.1) 1/9

Scale	$(m, n, p)$	Score Index (SI)
Absolutely more significance (AMS)	(0.9, 0.1, 0.1)	9
Very high significance (VS)	(0.8, 0.2, 0.2)	7
High significance (HS)	(0.7, 0.3, 0.3)	5
Slightly more significance (SMS)	(0.6, 0.4, 0.4)	3
Equally significance (ES)	(0.5, 0.5, 0.5)	1
Slightly significance (SS)	(0.4, 0.6, 0.4)	1/3
Low significance (LS)	(0.3, 0.7, 0.3)	1/5
Very low significance (VLS)	(0.2, 0.8, 0.2)	1/7
Absolutely low significance (ALS)	(0.1, 0.9, 0.1)	1/9

Step 2: Collection of expert evaluations using linguistic variables.

Experts express their opinions regarding the importance of each factor through linguistic assessments. These linguistic terms are defined using T-SFNs, which allow decision-makers to represent membership, abstinence, and non-membership degrees simultaneously. The linguistic scale used in this study is presented in Table 5. Let $T_{a} = (m_{a b}, n_{a b}, p_{a b})$ denote the evaluation of the a th specialists, where $a = 1, 2, \dots, k$ ; $b = 1, 2, \dots, l$ , along with the respective weights of k experts $w_{a} \in [0, 1]$ and $\sum_{a}^{k} w_{j} = 1$ .

Step 3: Aggregation of expert judgments.

The linguistic evaluations provided by the experts are converted into T-SFN. The aggregated significance value of each factor is calculated using the T-SFWAI operator in equation (25).

\begin{aligned} T - {SFWAI}_{w} \\ = (\sqrt[t]{(1 - \prod_{a = 1}^{k} {(1 - m_{a b}^{t})}^{a})}, \sqrt[t]{(\prod_{j = 1}^{k} {(1 - m_{a b}^{t})}^{w_{a}} - \prod_{j = 1}^{k} {(1 - m_{a b}^{t} - n_{a b}^{t})}^{w_{a}}),} \\ \sqrt[t]{(\prod_{a = 1}^{k} {(1 - m_{a b}^{t} - n_{a b}^{t})}^{w_{a}} - \prod_{a = 1}^{k} {(1 - m_{a b}^{t} - n_{a b}^{t} - v_{a b}^{t})}^{w_{b}})}) . \end{aligned}

(25)

The aggregation process combines individual expert judgments into a collective evaluation that reflects the overall importance of each factor.

Step 4: Defuzzify the aggregated indicators score.

After aggregating the expert evaluations, the T-SFN are converted into crisp values using the score function defined in equation (16). The score value represents the overall importance level of each factor and allows factors to be compared quantitatively.

Step 5: Determination of the threshold value.

In the final step, a threshold value is applied to determine whether a factor should be retained for further analysis. Equation (26) is employed to establish a threshold and verify the list of significant indicators.

D = \sum_{b = 1}^{l} \frac{d_{b}}{l},

(26)

where

d_{b}

denotes the score of the b th indicator. If

d_{b} < D

, then indicator

T_{b}

is removed. If

d_{b} > D

, then indicator

T_{b}

is confirmed.

3.4 T-Spherical Fuzzy DEMATEL (T-SF DEMATEL)

The procedure of the T-SF DEMATEL methodology is delineated as follows:

Step 1: Create direct influence matrices based on the expert's evaluation.

A linguistic scale in Table 6 is used to gather expert opinions and explain the impact of the criteria. Equation (15) formulates the direct effect matrix based on the pairwise comparisons $D^{a} = [m_{i j}, n_{i j}, p_{i j}]_{l \times l}$ .

Table 6.
Linguistic Scale for T-SF DEMATEL Method.

Scale $(m, n, p)$ Score Index (SI)

No Influence (NI) (0, 0.3, 0.2) 0

Weak Influence (WI) (0.35, 0.25, 0.25) 1

Moderate Influence (MI) (0.6, 0.2, 0.35) 2

Strong Influence (SI) (0.85, 0.15, 0.45) 3

Scale	$(m, n, p)$	Score Index (SI)
No Influence (NI)	(0, 0.3, 0.2)	0
Weak Influence (WI)	(0.35, 0.25, 0.25)	1
Moderate Influence (MI)	(0.6, 0.2, 0.35)	2
Strong Influence (SI)	(0.85, 0.15, 0.45)	3

Step 2: Combine direct influence into a matrix $(D^{agg})$ .

Equation (23) facilitates the creation of the aggregated direct influence matrix $(D^{agg})$ using the T-SFWAI technique for T-SF DEMATEL, as shown in equation (27).

(D^{a g g}) = [\begin{matrix} 0 & (m_{12}^{a g g}, n_{12}^{a g g}, p_{12}^{a g g}) & \dots & (m_{1 l}^{a g g}, n_{1 l}^{a g g}, p_{1 l}^{a g g}) \\ (m_{21}^{a g g}, n_{21}^{a g g}, p_{21}^{a g g}) & 0 & \dots & (m_{2 l}^{a g g}, n_{2 l}^{a g g}, p_{2 l}^{a g g}) \\ ⋮ & ⋮ & ⋱ & ⋮ \\ (m_{l 1}^{a g g}, n_{l 1}^{a g g}, p_{l 1}^{a g g}) & (m_{l 2}^{a g g}, n_{l 2}^{a g g}, p_{l 2}^{a g g}) & \dots & 0 \end{matrix}],

(27)

where

(m_{i j}^{a g g}, n_{i j}^{a g g}, p_{i j}^{a g g})

is the aggregated T-SF value of the impact of the ith to jth criteria.

Step 3: Formulate the initial direct influence matrix A.

A = s D; where s = m i n [\frac{1}{m a x_{i} \sum_{j = 1}^{l} | d_{i j} |}, \frac{1}{m a x_{j} \sum_{i = 1}^{l} | d_{i j} |}] .

(28)

The initial direct impact matrix X is formulated using equation (29), with s representing the normalization index.

\begin{array}{l} A^{m} = [\begin{matrix} 0 & m_{12} & \dots & m_{1 l} \\ m_{21} & 0 & \dots & m_{2 l} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ m_{l 1} & m_{l 2} & \dots & 0 \end{matrix}], A^{n} = [\begin{matrix} 0 & n_{12} & \dots & n_{1 l} \\ n_{21} & 0 & \dots & n_{2 l} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ n_{l 1} & n_{l 2} & \dots & 0 \end{matrix}], \\ A^{p} = [\begin{matrix} 0 & p_{12} & \dots & p_{1 l} \\ p_{21} & 0 & \dots & p_{2 l} \\ ⋮ & ⋮ & ⋱ & ⋮ \\ p_{1 l} & p_{l 2} & \dots & 0 \end{matrix}] . \end{array}

(29)

Step 4: Estimate the total impact matrix T utilizing equations (30) and (31):

\begin{aligned} T & = A + A^{'} = A (I - A) = [\begin{matrix} t_{11} & \dots & t_{1 l} \\ ⋮ & ⋱ & ⋮ \\ t_{l 1} & \dots & t_{l l} \end{matrix}], \end{aligned}

(30)

\begin{aligned} T & = [\begin{matrix} (m_{11}^{T}, n_{11}^{T}, p_{11}^{T}) & (m_{12}^{T}, n_{12}^{T}, p_{12}^{T}) & \dots & (m_{1 l}^{T}, n_{1 l}^{T}, p_{1 l}^{T}) \\ (m_{21}^{T}, n_{21}^{T}, p_{21}^{T}) & (m_{22}^{T}, n_{22}^{T}, p_{22}^{T}) & \dots & (m_{2 l}^{T}, n_{2 l}^{T}, p_{2 l}^{T}) \\ ⋮ & ⋮ & ⋱ & ⋮ \\ (m_{l 1}^{T}, n_{l 1}^{T}, p_{l 1}^{T}) & (m_{l 2}^{T}, n_{l 2}^{T}, p_{l 2}^{T}) & \dots & (m_{l l}^{T}, n_{l l}^{T}, p_{l l}^{T}) \end{matrix}] . \end{aligned}

(31)

Step 5: Calculate the T-SF column $(c o l_{j})$ and row sum $(r o w_{i})$ utilizing equations (32) and (33) as follows.

\begin{aligned} r o w_{i} & = \sum_{j = 1}^{l} (m_{i j}^{T}, n_{i j}^{T}, p_{i j}^{T}) i, j = 1, \dots, l, \end{aligned}

(32)

\begin{aligned} c o l_{j} & = \sum_{i = 1}^{l} (m_{i j}^{T}, n_{i j}^{T}, p_{i j}^{T}) i, j = 1, \dots, l . \end{aligned}

(33)

Step 6: Assess relationship importance and prominence.

Equation (16) converts defuzzification into score values. The $(r o w + c o l)$ value reflects indicator significance, while $(r o w - c o l)$ shows its causal role: positive for “cause,” negative for “effect.”

Step 7: Establish the Network Relations Map (NRM).

A causation and effect graph may be created by charting the dataset $(r o w + c o l, r o w - c o l \;)$ , with the horizontal axis from $(r o w + c o l)$ and the vertical axis from $(r o w - c o l)$ .

3.5 T-Spherical Fuzzy COCOSO (T-SF COCOSO)

Step 1: Construct a detailed evaluation matrix.

In this step, experts evaluate the effectiveness of each resilience strategy (Table 3) in addressing the identified resilience factors (Table 2) using the linguistic scale presented in Table 5. The evaluations from k experts on the effectiveness of $s$ strategies addressing l factors, weighted by $w_{a} \in [0, 1]$ and $\sum_{a = 1}^{k} w_{a} = 1$ are then combined into matrix T presented in equation (34) using the T-SFWAI operator in equation (23).

T = [\begin{matrix} t_{11} & \dots & t_{1 l} \\ ⋮ & ⋱ & ⋮ \\ t_{s 1} & \dots & t_{s l} \end{matrix}] .

(34)

Step 2: Assess the relative significance of the strategies.

The relative significance of the specified strategies $(i)$ is determined utilizing the Weighted Sum Method $W S_{i}$ and the Weighted Product Method $W P_{i}$ via equations (35) and (36), where $ω_{b} \in [0, 1]$ and $\sum_{b = 1}^{l} w_{b} = 1$ represents the crisp weights of l indicators, respectively:

\begin{aligned} W S_{i} & = \sum_{b = 1}^{l} t_{b i} \cdot ω_{b}, \end{aligned}

(35)

\begin{aligned} W P_{i} & = \sum_{b = 1}^{l} {t_{b i}}^{ω_{b}}, \end{aligned}

(36)

where

t_{b i} \cdot ω_{b}

is calculated by equation (20) and

{t_{b i}}^{ω_{b}}

is calculated by equation (21). The score function in equation (16) is applied to defuzzy

t_{b i} \cdot ω_{b}

and

{t_{b i}}^{ω_{b}}

before estimating

W S_{i}

and

W P_{i}

Step 3: Compute three appraisal scores, $k a$ , $k b$ and $k c$ using equations (37) to (39):

\begin{aligned} k a_{i} & = \frac{W S_{i} + W P_{i}}{\sum_{i = 1}^{s} (W S_{i} + W P_{i})}, \end{aligned}

(37)

\begin{aligned} k b_{i} & = \frac{1}{2} (\frac{W S_{i}}{M i n_{i} (W S_{i})} + \frac{W P_{i}}{M i n_{i} (W P_{i})}), \end{aligned}

(38)

\begin{aligned} k c_{i} & = \frac{α (W S_{i}) + (1 - α) (W P_{i})}{α Ma x_{i} (W S_{i}) + (1 - α) Ma x_{i} (W P_{i})}, 0 \leq α \leq 1, \end{aligned}

(39)

where

α

is assigned a value of 0.5.

Step 4: Rank the strategies.

The ultimate ranking of the options is established by organizing the $k_{i}$ values in descending order as expressed in equation (40).

k_{i} = \sqrt[3]{k a_{i} k b_{i} k c_{i}} + \frac{1}{3} (k a_{i} + k b_{i} + k c_{i}) .

(40)

4 Case Study

4.1 Experts Panel

The Delphi technique requires careful expert selection to ensure the validity and reliability of the results. In practice, a panel of approximately 10–20 well-informed specialists is generally considered appropriate for Delphi-based studies (Torres Vergara et al., 2024). In this study, a research council consisting of the authors was established to review the reputation of the candidate experts. A total of 20 professionals were selected, including seven academics, eight managers from CBEC firms, two policymakers, and three employees, with an average of 18.3 years of experience in the CBEC sector. The number of years of professional experience was used as the main criterion for assessing expertise. Specifically, the experts were evaluated using the linguistic scale presented in Table 7. These linguistic assessments were then converted into T-SFNs, aggregated using the T-SFWAI operator (equation (23)), and subsequently transformed into crisp values through the defuzzification formula (equation (16)). This procedure is consistent with recent T-spherical fuzzy decision-making frameworks, in which expert evaluations are represented through linguistic variables and transformed into T-SFN to capture uncertainty and improve the reliability of expert aggregation (Aytekin et al., 2024). The resulting expert weights are reported in Table 8.

Table 7.
Linguistic Scale for Evaluating Experts.

Experience in Years $(ζ)$ Expertise $(m, n, p)$

$ζ < 5$ Very Low (VL) (0.1, 0.1, 0.9)

$5 \leq ζ \leq 10$ Low (L) (0.2, 0.3, 0.9)

$10 \leq ζ \leq 15$ Medium Below (MB) (0.4, 0.2, 0.8)

$15 \leq ζ \leq 20$ Medium (M) (0.6, 0.2, 0.6)

$20 \leq ζ \leq 25$ Medium Above (MA) (0.8, 0.2, 0.5)

$25 \leq ζ \leq 30$ High (H) (0.9, 0.3, 0.2)

$ζ \geq 30$ Very High (VH) (0.9, 0.1, 0.1)

Experience in Years $(ζ)$	Expertise	$(m, n, p)$
$ζ < 5$	Very Low (VL)	(0.1, 0.1, 0.9)
$5 \leq ζ \leq 10$	Low (L)	(0.2, 0.3, 0.9)
$10 \leq ζ \leq 15$	Medium Below (MB)	(0.4, 0.2, 0.8)
$15 \leq ζ \leq 20$	Medium (M)	(0.6, 0.2, 0.6)
$20 \leq ζ \leq 25$	Medium Above (MA)	(0.8, 0.2, 0.5)
$25 \leq ζ \leq 30$	High (H)	(0.9, 0.3, 0.2)
$ζ \geq 30$	Very High (VH)	(0.9, 0.1, 0.1)

Table 8.

Demographics of CBEC Experts and Their Weights.

Experts	Jobs	Years	Expertise	Score	Weight	Experts	Jobs	Years	Expertise	Score	Weight
E1	Employee	12	MB	0.240	0.023	E11	Academician	15	M	0.480	0.046
E2	Manager	20	MA	0.675	0.064	E12	Academician	25	H	0.840	0.080
E3	Academician	14	MB	0.240	0.023	E13	Employee	14	MB	0.240	0.023
E4	Academician	17	M	0.480	0.046	E14	Manager	17	M	0.480	0.046
E5	Manager	22	MA	0.675	0.064	E15	Policy maker	21	MA	0.675	0.064
E6	Manager	16	M	0.480	0.046	E16	Policy maker	26	H	0.840	0.080
E7	Employee	12	MB	0.240	0.023	E17	Academician	24	MA	0.675	0.064
E8	Manager	21	MA	0.675	0.064	E18	Academician	18	M	0.480	0.046
E9	Academician	18	M	0.480	0.046	E19	Manager	19	M	0.480	0.046
E10	Manager	20	MA	0.675	0.064	E20	Manager	15	M	0.480	0.046

The conversion of experts’ years of experience into T-SFN prior to defuzzification aims to model the uncertainty associated with evaluating expert competence. In practice, years of experience only provide an approximate indication of an expert's knowledge or decision-making capability, and the relationship between experience and expertise is not necessarily linear. For example, the difference between nine and ten years of experience rarely leads to a meaningful difference in the quality of expert judgment. However, if raw numerical values of experience are directly used to assign weights, the model implicitly assumes that expertise increases linearly with each additional year. This assumption may exaggerate minor differences and introduce bias into the aggregation of expert opinions. To address this issue, years of experience are first transformed into T-SFNs, allowing the level of expertise to be represented through linguistic categories such as low, medium, and high while simultaneously capturing the ambiguity and hesitation in mapping experience to actual competence. Subsequently, a defuzzification process is applied to convert the fuzzy values into crisp scores, which are then used for normalization and for calculating the weights of experts in the aggregation procedures of the MCDM model. This approach enables the model to capture uncertainty during the expert evaluation stage while maintaining computational feasibility for subsequent analytical steps.

4.2 The T-SF Delphi Results

The T-SF Delphi method was applied to screen and validate the importance of the factors influencing the resilience of the CBECSC. In the first step, an extensive literature review was conducted to identify an initial set of 49 criteria, which served as the basis for constructing the expert evaluation questionnaire. In the second step, twenty experts evaluated the relevance and importance of these criteria using the linguistic scale presented in Table 5. The linguistic assessments provided by the experts were then converted into T-SF values to facilitate the Delphi computation. Subsequently, the evaluations from the twenty experts for each proposed factor were aggregated using the T-SF aggregation operator described in equation (25), where expert weights were determined based on Table 9. The aggregated scores were then compared with the predefined threshold value (D = 0.667) to determine whether each factor should be retained or eliminated from the analysis. The T-SF Delphi results presented in Table 9 indicate that three factors—F8 (Disaster Planning), F13 (Strategic Stock), and F42 (SC structure flexibility)—obtained scores below the threshold and were therefore removed from the final factor set. Most experts noted that these factors overlapped conceptually with other variables in the model. Specifically, F8 (Disaster Planning) is considered to be encompassed within F9 (Crisis Planning); F13 (Strategic Stock) is conceptually similar to F48 (Safety Stock Inventory); and F42 (SC Structure Flexibility) is represented through related flexibility factors such as F43 (Flexible Manufacturing System), F44 (Resource Flexibility), and F46 (Production Mix Flexibility).

Table 9.
The T-SF Delphi Results.

Factors Score Classify Factors Score Classify

F1 0.694 Confirmed F26 0.669 Confirmed

F2 0.680 Confirmed F27 0.676 Confirmed

F3 0.669 Confirmed F28 0.670 Confirmed

F4 0.701 Confirmed F29 0.688 Confirmed

F5 0.695 Confirmed F30 0.706 Confirmed

F6 0.688 Confirmed F31 0.688 Confirmed

F7 0.676 Confirmed F32 0.675 Confirmed

F8 0.397 Removed F33 0.677 Confirmed

F9 0.681 Confirmed F34 0.681 Confirmed

F10 0.686 Confirmed F35 0.698 Confirmed

F11 0.681 Confirmed F36 0.677 Confirmed

F12 0.678 Confirmed F37 0.680 Confirmed

F13 0.367 Removed F38 0.687 Confirmed

F14 0.693 Confirmed F39 0.698 Confirmed

F15 0.668 Confirmed F40 0.680 Confirmed

F16 0.670 Confirmed F41 0.699 Confirmed

F17 0.685 Confirmed F42 0.398 Removed

F18 0.685 Confirmed F43 0.679 Confirmed

F19 0.681 Confirmed F44 0.707 Confirmed

F20 0.689 Confirmed F45 0.711 Confirmed

F21 0.670 Confirmed F46 0.698 Confirmed

F22 0.668 Confirmed F47 0.699 Confirmed

F23 0.669 Confirmed F48 0.697 Confirmed

F24 0.682 Confirmed F49 0.703 Confirmed

F25 0.684 Confirmed $D$ 0 . 667

Factors	Score	Classify	Factors	Score	Classify
F1	0.694	Confirmed	F26	0.669	Confirmed
F2	0.680	Confirmed	F27	0.676	Confirmed
F3	0.669	Confirmed	F28	0.670	Confirmed
F4	0.701	Confirmed	F29	0.688	Confirmed
F5	0.695	Confirmed	F30	0.706	Confirmed
F6	0.688	Confirmed	F31	0.688	Confirmed
F7	0.676	Confirmed	F32	0.675	Confirmed
F8	0.397	Removed	F33	0.677	Confirmed
F9	0.681	Confirmed	F34	0.681	Confirmed
F10	0.686	Confirmed	F35	0.698	Confirmed
F11	0.681	Confirmed	F36	0.677	Confirmed
F12	0.678	Confirmed	F37	0.680	Confirmed
F13	0.367	Removed	F38	0.687	Confirmed
F14	0.693	Confirmed	F39	0.698	Confirmed
F15	0.668	Confirmed	F40	0.680	Confirmed
F16	0.670	Confirmed	F41	0.699	Confirmed
F17	0.685	Confirmed	F42	0.398	Removed
F18	0.685	Confirmed	F43	0.679	Confirmed
F19	0.681	Confirmed	F44	0.707	Confirmed
F20	0.689	Confirmed	F45	0.711	Confirmed
F21	0.670	Confirmed	F46	0.698	Confirmed
F22	0.668	Confirmed	F47	0.699	Confirmed
F23	0.669	Confirmed	F48	0.697	Confirmed
F24	0.682	Confirmed	F49	0.703	Confirmed
F25	0.684	Confirmed	$D$	0 . 667

4.3 The T-SF DEMATEL Results

Based on the T-SF Delphi results, T-SF DEMATEL analyzes the remaining 46 factors. The experts assessed each pair of elements using the linguistic scale in Table 5. Following a thorough methodology, Table 10 shows T-SF DEMATEL outcomes.

Table 10.
The T-SF DEMATEL Results.

Factors $r o w_{i}$ $c o l_{i}$ $r o w_{i} + c o l_{i}$ $r o w_{i} - c o l_{i}$ Classify Rank

F1 0.999999 0.999993 1.999992 0.0000060 Cause 14

F2 0.999996 0.999993 1.999989 0.0000031 Cause 30

F3 0.999997 0.999987 1.999984 0.0000105 Cause 43

F4 0.999999 0.999997 1.999996 0.0000020 Cause 5

F5 0.999997 0.999996 1.999993 0.0000016 Cause 12

F6 0.999998 0.999994 1.999992 0.0000034 Cause 16

F7 0.999997 0.999991 1.999988 0.0000051 Cause 36

F9 0.999995 0.999995 1.999990 −0.0000003 Effect 26

F10 0.999996 0.999994 1.999990 0.0000019 Cause 20

F11 0.999995 0.999994 1.999989 0.0000015 Cause 28

F12 0.999998 0.999990 1.999989 0.0000081 Cause 33

F14 0.999997 0.999996 1.999993 0.0000007 Cause 13

F15 0.999991 0.999986 1.999978 0.0000050 Cause 46

F16 0.999997 0.999989 1.999986 0.0000081 Cause 41

F17 0.999995 0.999995 1.999990 −0.0000001 Effect 22

F18 0.999999 0.999991 1.999990 0.0000074 Cause 21

F19 0.999996 0.999993 1.999990 0.0000027 Cause 27

F20 0.999997 0.999995 1.999992 0.0000028 Cause 15

F21 0.999993 0.999994 1.999987 −0.0000004 Effect 39

F22 0.999997 0.999984 1.999980 0.0000133 Cause 45

F23 0.999995 0.999991 1.999986 0.0000045 Cause 42

F24 0.999998 0.999992 1.999990 0.0000056 Cause 24

F25 0.999995 0.999995 1.999990 0.0000005 Cause 23

F26 0.999998 0.999985 1.999983 0.0000135 Cause 44

F27 0.999995 0.999993 1.999988 0.0000021 Cause 37

F28 0.999994 0.999992 1.999987 0.0000018 Cause 40

F29 0.999997 0.999995 1.999992 0.0000023 Cause 17

F30 0.999998 0.999999 1.999997 −0.0000004 Effect 3

F31 0.999995 0.999996 1.999991 −0.0000011 Effect 18

F32 0.999996 0.999992 1.999987 0.0000038 Cause 38

F33 0.999997 0.999991 1.999988 0.0000065 Cause 35

F34 0.999996 0.999994 1.999990 0.0000027 Cause 25

F35 0.999998 0.999996 1.999994 0.0000018 Cause 8

F36 0.999998 0.999990 1.999988 0.0000074 Cause 34

F37 0.999994 0.999995 1.999989 −0.0000019 Effect 31

F38 0.999998 0.999992 1.999990 0.0000063 Cause 19

F39 0.999998 0.999995 1.999994 0.0000033 Cause 10

F40 0.999993 0.999996 1.999989 −0.0000030 Effect 29

F41 0.999998 0.999998 1.999995 0.0000002 Cause 6

F43 0.999994 0.999994 1.999989 −0.0000002 Effect 32

F44 0.999998 0.999999 1.999997 −0.0000005 Effect 2

F45 0.999998 0.999999 1.999997 −0.0000006 Effect 1

F46 0.999994 0.999999 1.999994 −0.0000049 Effect 9

F47 0.999996 0.999999 1.999995 −0.0000028 Effect 7

F48 0.999999 0.999995 1.999993 0.0000042 Cause 11

F49 0.999998 0.999998 1.999996 −0.0000001 Effect 4

Factors	$r o w_{i}$	$c o l_{i}$	$r o w_{i} + c o l_{i}$	$r o w_{i} - c o l_{i}$	Classify	Rank
F1	0.999999	0.999993	1.999992	0.0000060	Cause	14
F2	0.999996	0.999993	1.999989	0.0000031	Cause	30
F3	0.999997	0.999987	1.999984	0.0000105	Cause	43
F4	0.999999	0.999997	1.999996	0.0000020	Cause	5
F5	0.999997	0.999996	1.999993	0.0000016	Cause	12
F6	0.999998	0.999994	1.999992	0.0000034	Cause	16
F7	0.999997	0.999991	1.999988	0.0000051	Cause	36
F9	0.999995	0.999995	1.999990	−0.0000003	Effect	26
F10	0.999996	0.999994	1.999990	0.0000019	Cause	20
F11	0.999995	0.999994	1.999989	0.0000015	Cause	28
F12	0.999998	0.999990	1.999989	0.0000081	Cause	33
F14	0.999997	0.999996	1.999993	0.0000007	Cause	13
F15	0.999991	0.999986	1.999978	0.0000050	Cause	46
F16	0.999997	0.999989	1.999986	0.0000081	Cause	41
F17	0.999995	0.999995	1.999990	−0.0000001	Effect	22
F18	0.999999	0.999991	1.999990	0.0000074	Cause	21
F19	0.999996	0.999993	1.999990	0.0000027	Cause	27
F20	0.999997	0.999995	1.999992	0.0000028	Cause	15
F21	0.999993	0.999994	1.999987	−0.0000004	Effect	39
F22	0.999997	0.999984	1.999980	0.0000133	Cause	45
F23	0.999995	0.999991	1.999986	0.0000045	Cause	42
F24	0.999998	0.999992	1.999990	0.0000056	Cause	24
F25	0.999995	0.999995	1.999990	0.0000005	Cause	23
F26	0.999998	0.999985	1.999983	0.0000135	Cause	44
F27	0.999995	0.999993	1.999988	0.0000021	Cause	37
F28	0.999994	0.999992	1.999987	0.0000018	Cause	40
F29	0.999997	0.999995	1.999992	0.0000023	Cause	17
F30	0.999998	0.999999	1.999997	−0.0000004	Effect	3
F31	0.999995	0.999996	1.999991	−0.0000011	Effect	18
F32	0.999996	0.999992	1.999987	0.0000038	Cause	38
F33	0.999997	0.999991	1.999988	0.0000065	Cause	35
F34	0.999996	0.999994	1.999990	0.0000027	Cause	25
F35	0.999998	0.999996	1.999994	0.0000018	Cause	8
F36	0.999998	0.999990	1.999988	0.0000074	Cause	34
F37	0.999994	0.999995	1.999989	−0.0000019	Effect	31
F38	0.999998	0.999992	1.999990	0.0000063	Cause	19
F39	0.999998	0.999995	1.999994	0.0000033	Cause	10
F40	0.999993	0.999996	1.999989	−0.0000030	Effect	29
F41	0.999998	0.999998	1.999995	0.0000002	Cause	6
F43	0.999994	0.999994	1.999989	−0.0000002	Effect	32
F44	0.999998	0.999999	1.999997	−0.0000005	Effect	2
F45	0.999998	0.999999	1.999997	−0.0000006	Effect	1
F46	0.999994	0.999999	1.999994	−0.0000049	Effect	9
F47	0.999996	0.999999	1.999995	−0.0000028	Effect	7
F48	0.999999	0.999995	1.999993	0.0000042	Cause	11
F49	0.999998	0.999998	1.999996	−0.0000001	Effect	4

The results show 33 causative factors and 13 effect factors: F9, F17, F21, F30, F31, F37, F40, F43, F44, F45, F46, F47, and F49. Most of the impacting aspects are in the Supply Chain Adaptability category, specifically the Flexibility and Redundancy sub-dimensions with F43, F44, F45, F46, F47, and F49. Flexibility and redundancy in the supply chain function as reactive mechanisms to address interruptions or hazards, rather than serving as primary motivators for change. Factors like F44, F46, and F47 are only effective when facilitated by strategic management factors such as F1 or F6. The results of this study differ from those of Liu et al. (2021) due to the research setting. Although Liu et al. (2021) found that the most effective factors relate to risk management culture and adaptive management, they also belong to Supply Chain Adaptability. Preventive and reactive methods affect flexibility and redundancy in Vietnam, depending on resource management and infrastructure development. However, China's supply chain has developed quickly and is resilient to disruptions.

Table 10 also reveals the ranking of the $(r o w_{i} + c o l_{i})$ values, indicating that F45, F44, and F30 are the top three with the highest $(r o w_{i} + c o l_{i})$ values.

F45 has the highest value at 1.999997. Many Vietnamese small and medium companies lack supply chain management skills, requiring an emphasis on understanding and applying these techniques. Thus, Supply Chain management knowledge is essential for resilience.

The second highest $(r o w_{i} + c o l_{i})$ rating of F44 highlights the strategic importance of resource flexibility in synchronizing supply chain components. This is essential for adapting and stabilizing Vietnam's volatile supply chain.

The third highest $(r o w_{i} + c o l_{i})$ rating of F30 underscores the crucial function of collaborative planning within Vietnam's supply chain. This is the key to connecting and coordinating activities and ensuring system partner collaboration.

These characteristics are critical to CBECSC resilience, although other parameters influence them and can be indirectly affected by causal factors.

An Influence Relationship Map (IRM) was created. The horizontal axis is the line passing through point $(r o w_{i} - c o l_{i}) = 0$ to prioritize factors. Note that the vertical axis passes through point $(r o w_{i} + c o l_{i}) = 1.9999973$ (the average of all values). The IRM has four quadrants as indicated in Figure 1.

Figure 1.

IRM of CBECSC resilience in Vietnam.

Quadrant I: Comprises factors characterized by significant relevance and minimal prominence. The factors within this category exhibit a significant degree of autonomy, influencing the other factors more than they are influenced themselves. These factors are deemed the primary reason and require the utmost attention.

Quadrant II: The factors in the second quadrant exhibit a robust correlation and significance. These factors warrant considerable consideration, as they constitute the leading causes and substantially influence the other factors.

Quadrant III: Indicators have little correlation and significance. The system is little affected by these separate influences. Group III elements need the least attention of the four categories.

Quadrant IV: Factors in this group have modest associations but high relevance. Thus, systemic influences greatly affect these characteristics. DMs can improve group IV elements by enhancing the elements from Quadrants I and II, as these elements influence the effect of group IV.

Table 11 shows factor priorities. Prioritized CBECSC resilience components in Vietnam showcase Vietnam's economic and logistical uniqueness. Quadrant I variables are core and require speedy decision-making, according to Table 11. The survey lists Vietnam's top CBECSC resilience factors as F18, F38, F1, F48, and F6. These factors reflect the export sectors, economic policies, and supply chain issues.

Table 11.

The Priority of Factor.

Quad.	Factors	Classify	Order Priority
I	F18, F38, F1, F48, F6, F39, F20, F29, F35, F10, F5, F14, F41,F4.	Core Factor	First
II	F26, F22, F3, F12, F16, F36, F33, F24, F7, F23, F32, F2, F19, F2, F34, F27, F28, F11, F25.	Driving Factor	Second
III	F43, F9, F21, F37, F40.	Independent Factor	Fourth
IV	F17, F49, F31,F30, F44, F45, F47, F46.	Impact Factor	Third

4.4 The T-SF COCOSO Results

Experts assess six resilience alternatives for CBECSC. Table 5 lists language expert assessments. These evaluations are converted to T-SF numbers and evaluated using T-SF COCOSO. Table 12 presents the T-SF COCOSO results with $α = 0.5$ . T-SF COCOSO ranks ten potential CBECSC resilience improvements in Vietnam. Top strategies include A3, A4, and A2, as listed in Table 12. These approaches are compared to previous research to contextualize their relevance and usefulness. The T-SF COCOSO results emphasize a clear prioritization, with strategies ranked as follows: A3 > A4 > A2 > A6 > A1 > A5. Decision-makers are advised to adopt this order when formulating initiatives to enhance CBECSC resilience.

Table 12.
The T-SF COCOSO Results.

Stra. $k a_{i}$ $k b_{i}$ $k c_{i}$ $k_{i}$ Rank

A1 0.1657 3.0135 0.9811 2.1750 5

A2 0.1673 3.6928 0.9908 2.4660 3

A3 0.1689 4.9697 1.0000 2.9894 1

A4 0.1688 4.8586 0.9997 2.9450 2

A5 0.1625 2.0000 0.9625 1.7206 6

A6 0.1668 3.5814 0.9881 2.4177 4

Stra.	$k a_{i}$	$k b_{i}$	$k c_{i}$	$k_{i}$	Rank
A1	0.1657	3.0135	0.9811	2.1750	5
A2	0.1673	3.6928	0.9908	2.4660	3
A3	0.1689	4.9697	1.0000	2.9894	1
A4	0.1688	4.8586	0.9997	2.9450	2
A5	0.1625	2.0000	0.9625	1.7206	6
A6	0.1668	3.5814	0.9881	2.4177	4

5 Discussion

5.1 Discussion

This study employs the T-SF DEMATEL method to investigate the causal factors influencing CBECSC resilience in Vietnam. The results list Vietnam's top CBECSC resilience factors as F18, F38, F1, F48, and F6. These factors reflect the export sectors, economic policies, and supply chain issues.

F18 - Customize Products

The Vietnamese textile and agriculture industries depend on product modification to meet global market demands for customized designs and quality certifications. Textile customization boosts value and competitiveness. International standards for post-harvest management and packaging enhance the quality of agricultural products. Unlike Liu et al. (2021) and Nabil et al. (2024) which emphasize standardization and mass production in global supply chains, this study underscores the strategic advantage of customization in Vietnam's context. These findings highlight Vietnam's opportunity to differentiate its products in competitive markets, particularly against China and India.

F38 - Mutual Contingency Plan

Mutual contingency planning enables firms to coordinate disruption-response mechanisms and share resources to mitigate supply chain disruptions more effectively. In the context of Vietnam's fragmented logistics infrastructure and the high exposure of CBECSC to external shocks, this factor plays an important role in strengthening the resilience of the CBECSC. Specifically, mutual contingency planning allows firms to coordinate emergency responses, exchange critical logistics information, and mobilize alternative suppliers or transport resources when disruptions occur, thereby reducing recovery time and maintaining operational continuity. Previous studies have emphasized the importance of collaboration and information sharing in improving supply chain resilience. For example, Nabil et al. (2024) highlights the role of inter-organizational cooperation and knowledge sharing among supply chain partners. However, the present study extends this perspective by emphasizing structured mutual contingency planning, where firms proactively establish coordinated disruption-response mechanisms rather than relying solely on informal collaboration.

F1 - Trade Policies

Customs regulations, fiscal incentives, and government policy support play an important role in improving supply chain performance and enabling access to international markets. Streamlined trade policies can reduce operational costs while enhancing logistics efficiency. Zhang et al. (2023) note that well-designed trade policies help mitigate supply chain disruptions and lower operational expenses. This study further explores how policy frameworks interact with Vietnam's dependence on imports and its export-oriented logistics infrastructure. Within the Vietnamese context, trade policies are especially critical because the country's CBEC ecosystem relies heavily on import flows, export-driven production, and international logistics networks. Although Vietnam has made notable progress in trade liberalization and digital trade facilitation, several challenges remain. Issues such as inconsistent customs procedures, fragmented regulatory systems, and varying documentation requirements continue to hinder the efficiency of cross-border logistics operations. Addressing these structural barriers is therefore essential for strengthening supply chain resilience. From a policy standpoint, several practical measures can enhance the effectiveness of trade policies in supporting CBECSC. First, policymakers may further promote the digitalization of customs clearance processes and expand national single-window systems to shorten border processing times. Second, targeted fiscal incentives and supportive regulatory programs could encourage SMEs to adopt resilience-oriented logistics practices, such as supplier diversification and safety stock management. Third, enhancing cross-border regulatory collaboration and aligning logistics standards with major trading partners may further improve the stability and operational efficiency of Vietnam's CBECSC.

F48 - Safety Stock Inventory

Safety stocks protect against natural disasters, logistical issues, and variations in market demand. This method is crucial for Vietnam due to its significant exposure to risk. Safety stock inventory is less frequently mentioned in Liu et al. (2021) and Nabil et al. (2024), suggesting a distinct focus in Vietnam, where operational stability depends on maintaining adequate inventory reserves.

F6 - Reconfigure Contingency Strategy

Vietnam's turbulent market requires dynamic emergency response solutions to handle unexpected interruptions. This adaptability enables firms to utilize resources effectively and survive disasters. In contrast to Liu et al. (2021), this study prioritizes reconfiguration, reflecting Vietnam's need for flexible and proactive risk management.

In conclusion, this analysis prioritizes CBECSC resilience elements similarly to and differently from past research. First, trade policies impact operating costs and supply chain stability, aligning with the resilience focus of Zhang et al.'s (2023) study. Secondly, unlike these studies, Liu et al. (2021) and Nabil et al. (2024), which prioritize factors such as partner cooperation and risk management, this study identifies product customization, safety stock inventory, and contingency strategy reconfiguration as distinctively relevant to Vietnam. These discrepancies highlight Vietnam's export-oriented approach and the challenges of competing with larger economies.

Speed of recovery, a prominent component in previous studies, is ranked 11th in this study, showing its importance but lesser priority than proactive resilience measures in Vietnam. This result contradicts earlier research. The speed of recovery, partner cooperation, and risk management all affect CBECSC resilience, according to Liu et al. (2021) and Nabil et al. (2024). According to the Vietnam study, speed of recuperation is a Quadrant I factor but rates 11th in priority. This report also highlights the characteristics of the Vietnamese market, including product customization, trade policies, and safety stock inventory. In rivalry with China and India, Vietnam is increasing its production scale but falling short in terms of cost and efficiency. Vietnamese supply chains are expensive and time-consuming due to the need to import raw materials and accessories. Product customization helps Vietnam compete globally. E-commerce companies in Vietnam can differentiate themselves and enhance competitiveness by offering flexibility, value, and catering to diverse client needs. In CBECSC, trade regulations affect operational costs, supply chain stability, and corporate competitiveness. Similar results were found by Zhang et al. (2023).

Moreover, this study ranks the CBECSC resilience strategies by using the T-SF COCOSO method. The order discovered is A3 > A4 > A2 > A6 > A1 > A5.

A3 Implementing Contingency Plans and Backup Plans (Rank 1)

This strategy focuses on establishing comprehensive contingency and backup plans to manage supply chain disruptions effectively. It particularly highlights the importance of factors F48, F6, and F38. In the Vietnamese context, many SMEs lack well-defined response mechanisms for dealing with unexpected disruptions, making such contingency strategies especially critical. Although Liu et al. (2021) did not specifically examine SME implementation, their study emphasizes partner collaboration and crisis preparedness as fundamental elements of supply chain resilience. Furthermore, this research underscores the constraints posed by Vietnam's limited transportation infrastructure and resource availability. While Zhang et al. (2023) primarily highlighted technological solutions for improving supply chain performance, the present study places greater emphasis on contingency strategies tailored to Vietnam's infrastructure limitations and resource conditions. Nevertheless, the implementation of contingency planning may present financial and operational challenges for SMEs. Establishing alternative suppliers, maintaining buffer inventories, or investing in monitoring technologies often requires additional financial resources and managerial capabilities that smaller firms may lack. As a result, policy support mechanisms, collaborative logistics arrangements, and phased implementation approaches may be necessary to enable SMEs to adopt contingency planning practices more effectively.

A4 Promoting Collaboration and Fairness (Rank 2)

Domestic and international supply chain partners must collaborate to develop trust and alignment. This approach addresses transparency and strategy alignment in F41 and F39. Nabil et al. (2024) emphasized the importance of partner cooperation but overlooked the importance of justice and transparency, which are essential to Vietnam's economic environment. This study emphasizes fairness as a cornerstone of successful teamwork.

A2 Speeding Up Recovery (Rank 3)

Supply chain disruptions must be resolved quickly in competitive e-commerce. This approach stresses F4 and F5. This report indicates that Vietnam's insufficient use of real-time monitoring and decision-making systems hinders recovery, which is crucial. The speed of recovery is a widely recognized factor, as noted in studies such as Liu et al. (2021) However, in this analysis, it ranks third in Vietnam due to infrastructure and technological constraints.

A6 Building risk management team (Rank 4)

Reducing supply chain vulnerabilities requires qualified risk management teams. This technique targets F20 and F39. Building such teams takes time and resources, reducing their usefulness. Research by Zhang et al. (2023) did not expressly prioritize the formation of a risk management team, underlining its special importance to Vietnam's SME-dominated market.

A1 Government Support (Rank 5)

Tax incentives, infrastructure development, and trade policy have a significant impact on supply chains. This strategy stresses F1. Although government backing is crucial, delaying policy implementation typically lessens its immediate influence. This study stated that Zhang et al. (2023) similarly identified trade policies but did not address policy execution inefficiencies.

A5 Strengthening International Coordination (Rank 6)

Strategically strengthening international coordination enables Vietnam to integrate more effectively into global production systems. This approach stresses F44 and F30. Differences in national norms limit its direct relevance to current situations. This study's findings support Nabil et al. (2024) emphasis on strategic international coordination.

5.2 Comparative Analysis

This study assesses the reliability of the method using T-SF MARCOS and T-SF TOPSIS ranking systems. TOPSIS and MARCOS assess the distance to the positive and negative ideal solutions (PIS and NIS) for finding the best alternative. The TOPSIS optimal solution is farthest from the NIS and closest to the PIS. In addition to the PIS and NIS, MARCOS evaluates the solution against an intermediary reference point, boosting stability. COCOSO, like TOPSIS and MARCOS, is independent of PIS and NIS distance. Instead, it combines ranking methods to balance criteria. Instead of using distance alone, COCOSO lets decision-makers quickly modify criteria weights using multiple combination strategies. Since TOPSIS and MARCOS share the same logic but differ from COCOSO, this study examines whether the results remain consistent or fluctuate using different methods. The approaches use different conceptual frameworks for calculations, but the solution rankings are the same. The reliability of the method is excellent. Table 13 shows T-SF MARCOS results, while Table 14 shows T-SF TOPSIS results.

Table 13.
T-SF TOPSIS Results.

Strategies D( $X_{i}$ , $X^{-}$ ) D( $X_{i}$ , $X *$ ) Closeness Coefficient Rank

A1 0.066 0.061 0.5201 5

A2 0.069 0.059 0.5407 3

A3 0.074 0.058 0.5629 1

A4 0.071 0.060 0.5426 2

A5 0.056 0.079 0.415 6

A6 0.066 0.060 0.5247 4

Strategies	D( $X_{i}$ , $X^{-}$ )	D( $X_{i}$ , $X *$ )	Closeness Coefficient	Rank
A1	0.066	0.061	0.5201	5
A2	0.069	0.059	0.5407	3
A3	0.074	0.058	0.5629	1
A4	0.071	0.060	0.5426	2
A5	0.056	0.079	0.415	6
A6	0.066	0.060	0.5247	4

Table 14.

T-SF MARCOS Results.

Strategies	$K_{i}^{-}$	$K_{i}^{+}$	$f (K_{i}^{-})$	$f (K_{i}^{+})$	$f (K_{i})$	Rank
A1	1.1124	0.9044	0.5516	0.4484	0.6628	5
A2	1.1221	0.9122	0.5516	0.4484	0.6685	3
A3	1.1328	0.9210	0.5516	0.4484	0.6749	1
A4	1.1303	0.9189	0.5516	0.4484	0.6734	2
A5	1.0898	0.8860	0.5516	0.4484	0.6493	6
A6	1.1195	0.9101	0.5516	0.4484	0.6670	4

Figure 2 depicts the comparative results of the three methods. Based on the ranking results in Tables 12 and 13, the strategies have the following consistent order: A3 > A4 > A2 > A6 > A1 > A5. This result is strong evidence that confirms the reliability of the proposed model.

Figure 2.

Comparative analysis result.

6 Conclusion

6.1 Conclusion

Using T-SF MARCOS and T-SF TOPSIS ranking systems, this study investigates method dependability. To select the optimum option, TOPSIS and MARCOS compare the distance to PIS and NIS. TOPSIS’ optimal solution is the farthest from NIS and the closest to PIS. In addition to the PIS and NIS, MARCOS analyzes the solution against an intermediary reference point, increasing stability. Like TOPSIS and MARCOS, COCOSO is PIS and NIS distance independent. Ranking systems are used to balance criteria.

Instead of using distance alone, COCOSO enables decision-makers to change criteria weights using numerous combination tactics easily. TOPSIS and MARCOS employ the same logic but differ from COCOSO; therefore, this study evaluates whether the findings are consistent or fluctuate across different methodologies. It helps evaluate the method's reliability. Different conceptual frameworks are used for calculations, but the rankings of solutions remain the same.

6.2 Theoretical Implications

This research extends the CBECSC resilience theory and complex system decision-making framework. The findings and techniques complement academic discourse in the following ways. The T-SF theory is utilized in conjunction with the Delphi, DEMATEL, and COCOSO MCDM methods to enhance its theoretical applications. The integration reveals that T-SF theory may address decision-making ambiguity, reluctance, and interdependencies that Fuzzy and MCDM methods miss. T-SFS clarify resilience factor analysis and reveal causal hierarchies. The T-SF DEMATEL method organizes 46 proven factors into cause-and-effect groups to systematically and scalably evaluate CBECSC resilience. By theoretically analyzing the interactions of resilience factors, researchers can identify underlying reasons and high-priority intervention areas. T-SF COCOSO's ability to prioritize solutions based on multiple factors yields a replicable model for complex decision-making, furthering decision science. This paper fills an important gap by examining Vietnam's SCRE, an area that has been explored mainly in developed markets. In emerging economies, distinct logistical and regulatory contexts make customization, collaboration, and trade policy especially critical. Our findings situate CBECSC resilience theory within this context, highlighting how contingency planning and rapid recovery can help mitigate disruptions in CBEC. Incorporating expert-weighted evaluations enhances methodological rigor while capturing diverse theoretical perspectives. The study employs fuzzy decision-making with T-SFS to account for membership, non-membership, and hesitancy, thereby enhancing the modeling of dynamic uncertainty and stakeholder preferences. By comparing MCDM methods such as T-SF MARCOS and T-SF TOPSIS, the research ensures methodological consistency and contributes to the broader decision science literature.

6.3 Practical Implications

This paper guides Vietnamese businesses, governments, and supply chains in enhancing their resilience to CBECSC. Organizations should use safety stock and alternative logistics paths to avoid disruptions. Given Vietnam's infrastructure challenges and reliance on SMEs, which lack reaction systems, these initiatives are crucial. Domestic and foreign partners are needed to improve transparency, strategy, and supply chain confidence. Governments streamline trade policies, optimize customs operations, and encourage investments in logistical infrastructure to enhance supply chain efficiency. The study also recommends government-supported training to help SMEs build professional risk management teams and implement proactive resilience measures. Promote resource flexibility, collaborative planning, and product customization to complement Vietnam's textiles and agriculture capabilities in supply chain strategies. Real-time monitoring and decision-making with digital technology speeds recovery and client confidence during disruptions. These practical tips guide stakeholders through Vietnam's changing e-commerce ecosystem and position its supply chains for success.

6.4 Limitations and Future Work

This study sheds light on CBECSC resiliency in Vietnam, although it has limitations. First, this research, which focuses on Vietnam, limits the applicability of the findings to other regions with different economic, regulatory, and logistical situations. To assess its flexibility and applicability, future studies should apply the paradigm to other emerging markets. Second, the study only covers the T-SF theory and the chosen MCDM approaches, neglecting advanced Fuzzy Set Theories, such as q-Rung Orthopair or Neutrosophic Fuzzy Sets, which could improve resilience modeling precision. Third, even with weighted ratings, expert opinions are subjective. Real-world performance measures could validate and improve the recommended solutions. The study also examines DEMATEL, COCOSO, and MARCOS MCDM approaches, facilitating future investment decisions.

Footnotes

Acknowledgements

The authors would like to express their heartfelt appreciation and gratitude to stakeholders for their contributions to this study.

ORCID iDs

Ming-Hua Lin

Phi-Hung Nguyen

Ethical Approval

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of FPT University (approval number: 2025/DHFPT).

Author Contributions

Conceptualization: J.-F.T., T.-H.T., and P.-H.N.; Methodology: J.-F.T., T.-H.T., and P.-H.N.; Software: T.-H.T. and P.-H.N.; Validation: J.-F.T., M.-H.L., and Z.-H.C.; Formal analysis: J.-F.T., M.-H.L., and Z.-H.C.; Investigation: T.-H.T. and P.-H.N.; Resources: J.- F.T., T.-H.T., and P.-H.N.; Data curation: T.-H.T. and P.-H.N.; Writing–original draft: T.-H.T. and P.-H.N.; Writing–review & editing: J.-F.T., T.-H.T., P.-H.N., Z.-H.C., and M.-H.L.; Visualization: T.-H.T.; Supervision: J.-F.T. and M.-H.L.; Project administration: J.-F.T.; Funding acquisition: J.-F.T.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

References

Anjum

Simic

Alrasheedi

M. A.

Shahab

(2024). T-spherical fuzzy-CRITIC-WASPAS model for the evaluation of cooperative intelligent transportation system scenarios. IEEE Access, 12, 61137–61151. https://doi.org/10.1109/ACCESS.2024.3392019

Atanassov

(1986). Intuitionistic fuzzy sets. Fuzzy Sets and Systems, 20(1), 87–96. https://doi.org/10.1016/S0165-0114(86)80034-3

Aytekin

Korucuk

Bedirhanoğlu

Ş. B.

Simic

(2024). Selecting the ideal sustainable green strategy for logistics companies using a T-spherical fuzzy-based methodology. Engineering Applications of Artificial Intelligence, 127(B), 107347. https://doi.org/10.1016/j.engappai.2023.107347

Ayyildiz

(2021). Interval valued intuitionistic fuzzy analytic hierarchy process-based green supply chain resilience evaluation methodology in post COVID-19 era. Environmental Science and Pollution Research, 30(15), 42476–42494. https://doi.org/10.1007/s11356-021-16972-y

Banomyong

Bhusiri

Julagasigorn

Varadejsatitwong

(2025). Assessing supply chain resilience to mitigate disruption: The focus on cross-border suppliers. Logistics, 9(1), 1. https://doi.org/10.3390/logistics9010001

Beikkhakhian

Javanmardi

Karbasian

Khayambashi

(2015). The application of ISM model in evaluating agile suppliers selection criteria and ranking suppliers using fuzzy TOPSIS-AHP methods. Expert Systems with Applications, 42(15-16), 6224–6236. https://doi.org/10.1016/j.eswa.2015.02.035

Chen

Sohal

A. S.

Prajogo

D. I.

(2013). Supply chain operational risk mitigation: A collaborative approach. International Journal of Production Research, 51(7), 2186–2199. https://doi.org/10.1080/00207543.2012.727490

Chen

Q. Y.

Liu

H. C.

Wang

J. H.

Shi

(2022). New model for occupational health and safety risk assessment based on fermatean fuzzy linguistic sets and CoCoSo approach. Applied Soft Computing, 126, 109262. https://doi.org/10.1016/j.asoc.2022.109262

Chen

Qiu

Wang

Yang

(2022). Analysis of effects on the dual circulation promotion policy for cross-border E-commerce B2B export trade based on system dynamics during COVID-19. Systems, 10(1), 13. https://doi.org/10.3390/systems10010013

10.

Chowdhury

M. M. H.

Quaddus

(2017). Supply chain resilience: Conceptualization and scale development using dynamic capability theory. International Journal of Production Economics, 188, 185–204. https://doi.org/10.1016/j.ijpe.2017.03.020

11.

Cohen

Cui

Doetsch

Ernst

Huchzermeier

Kouvelis

Lee

Matsuo

Tsay

A. A.

(2022). Bespoke supply-chain resilience: The gap between theory and practice. Journal of Operations Management, 68(5), 515–531. https://doi.org/10.1002/joom.1184

12.

Cui

Liu

Rani

Alrasheedi

(2021). Internet of Things (IoT) adoption barriers for the circular economy using Pythagorean fuzzy SWARA-CoCoSo decision-making approach in the manufacturing sector. Technological Forecasting and Social Change, 171, 120951. https://doi.org/10.1016/j.techfore.2021.120951

13.

Cuong

B. C.

(2014). Picture fuzzy sets. J Comput Sci Cybern, 30(4), 409–420. https://doi.org/10.15625/1813-9663/30/4/5032

14.

Das

Datta

Kumar

Kazancoglu

Ram

(2022). Building supply chain resilience in the era of COVID-19: An AHP-DEMATEL approach. Operations Management Research, 15(1-2), 249–267. https://doi.org/10.1007/s12063-021-00200-4

15.

Dohale

Gunasekaran

Akarte

Verma

(2021). An integrated Delphi-MCDM-Bayesian Network framework for production system selection. International Journal of Production Economics, 242, 108296. https://doi.org/10.1016/j.ijpe.2021.108296

16.

Er-retby

Mghazli

M. O.

Bahrar

El Mankibi

Benzaazoua

(2025). Assessment of building performance indicators using fuzzy DEMATEL approach: Comparing early-career and mid-expert perspectives. Journal of Building Engineering, 114, 114241. https://doi.org/10.1016/j.jobe.2025.114241

17.

Fetanat

Tayebi

(2024). Sustainability and availability prioritization of solar thermal energy technologies for enhanced oil recovery using T-spherical fuzzy set-based ordinal priority approach. Clean Technologies and Environmental Policy, 27(6), 2585–2614. https://doi.org/10.1007/s10098-024-03047-y

18.

Garola

Siragusa

Seghezzi

Mangiaracina

(2023). Managing COVID-19 disruption: The response of express couriers and lessons learned to improve resilience. The International Journal of Logistics Management, 34(7), 121–141. https://doi.org/10.1108/IJLM-03-2022-0114

19.

Giuffrida

Mangiaracina

Perego

Tumino

(2017). Cross-border B2C e-commerce to Greater China and the role of logistics: A literature review. International Journal of Physical Distribution & Logistics Management, 47(9), 772–795. https://doi.org/10.1108/IJPDLM-08-2016-0241

20.

Görçün

ÖF

Chatterjee

Stević

Küçükönder

(2024). An integrated model for road freight transport firm selection in third-party logistics using T-spherical fuzzy sets. Transp Res Part E Logist Transp Rev, 186, 103542. https://doi.org/10.1016/j.tre.2024.103542

21.

Govindan

Khodaverdi

Vafadarnikjoo

(2015). Intuitionistic fuzzy based DEMATEL method for developing green practices and performances in a green supply chain. Expert Systems with Applications, 42(20), 7207–7220. https://doi.org/10.1016/j.eswa.2015.04.030

22.

Gündoğdu

F. K.

(2019). Principals of spherical fuzzy sets. In International conference on intelligent and fuzzy systems (pp. 15–23). Springer.

23.

Guo

Jin

Zhao

(2023). E-commerce and supply chain resilience during COVID-19: Evidence from agricultural input e-stores in China. Journal of Agricultural Economics, 74(2), 369–393. https://doi.org/10.1111/1477-9552.12518

24.

Gupta

(2026). Enhancing freight logistics: Overcoming barriers and driving sustainable growth with an integrated grey Delphi DEMATEL approach. Benchmarking: An International Journal, 33(2), 368–391. https://doi.org/10.1108/BIJ-03-2024-0201

25.

Haldar

Ray

Banerjee

Ghosh

(2012). A hybrid MCDM model for resilient supplier selection. International Journal of Management Science and Engineering Management, 7(4), 284–292. https://doi.org/10.1080/17509653.2012.10671234

26.

Ivanov

(2020). Viable supply chain model: Integrating agility, resilience and sustainability perspectives—lessons from and thinking beyond the COVID-19 pandemic. Annals of Operations Research, 319(1), 1411–1431. https://doi.org/10.1007/s10479-020-03640-6

27.

Jafarzadeh Ghoushchi

Khazaeili

Amini

Osgooei

(2019). Multi-criteria sustainable supplier selection using piecewise linear value function and fuzzy best-worst method. Journal of Intelligent & Fuzzy Systems, 37(2), 2309–2325. https://doi.org/10.3233/JIFS-182609

28.

Liang

Luo

Dong

Santibanez Gonzalez

E.D.R.

Wang

(2021). T-spherical fuzzy TODIM method for multi-criteria group decision-making problem with incomplete weight information. Soft Computing, 25, 2981–3001. https://doi.org/10.1007/s00500-020-05357-x

29.

Kaya

Çolak

Terzi

(2018). Use of MCDM techniques for energy policy and decision-making problems: A review. International Journal of Energy Research, 42(7), 2344–2372. https://doi.org/10.1002/er.4016

30.

Kessler

Arlinghaus

J. C.

(2022). Managing supply chain disruption by collaborative resource sharing. In Dolgui

Ivanov

Sokolov

(Eds.), Supply network dynamics and control (pp. 79–93). Springer.

31.

Kutlu Gündoğdu

Kahraman

(2019). Spherical fuzzy sets and spherical fuzzy TOPSIS method. Journal of Intelligent & Fuzzy Systems, 36, 337–352. https://doi.org/10.3233/JIFS-181401

32.

Kutlu Gündoğdu

Kahraman

(2020). A novel spherical fuzzy analytic hierarchy process and its renewable energy application. Soft Computing, 24(6), 4607–4621. https://doi.org/10.1007/s00500-019-04222-w

33.

Liu

Chen

(2023). Enablers for maritime supply chain resilience during pandemic: An integrated MCDM approach. Transportation Research Part A: Policy and Practice, 175, 103777. https://doi.org/10.1016/j.tra.2023.103777

34.

Liu

Dou

Yang

(2021). Research on influencing factors of cross border E-Commerce supply chain resilience based on integrated fuzzy DEMATEL-ISM. IEEE Access, 9, 36140–36153. https://doi.org/10.1109/ACCESS.2021.3059867

35.

López

Ishizaka

Gul

Yücesan

Valencia

(2022). A calibrated fuzzy best-worst-method to reinforce supply chain resilience during the COVID 19 pandemic. Journal of the Operational Research Society, 74(9), 1968–1991. https://doi.org/10.1080/01605682.2022.2122739

36.

M. T.

Hsu

C. C.

Liou

J. J. H.

H.-W.

(2018). A hybrid MCDM and sustainability-balanced scorecard model to establish sustainable performance evaluation for international airports. Journal of Air Transport Management, 71, 9–19. https://doi.org/10.1016/j.jairtraman.2018.05.008

37.

Liu

Feng

Zhang

(2025). A system dynamics approach to resilience analysis in the sino-Russian timber supply chain. Forests, 16(7), 1106. https://doi.org/10.3390/f16071106

38.

Mahmood

Ullah

Khan

Jan

(2019). An approach toward decision-making and medical diagnosis problems using the concept of spherical fuzzy sets. Neural Computing and Applications, 31(11), 7041–7053. https://doi.org/10.1007/s00521-018-3521-2

39.

Ministry of Industry and Trade of Vietnam. (2023, 25 July). Vietnam E-commerce White Paper 2023.

40.

Mohammed

Harris

Soroka

Nujoom

(2019). A hybrid MCDM-fuzzy multi-objective programming approach for a G-resilient supply chain network design. Computers & Industrial Engineering, 127, 297–312. https://doi.org/10.1016/j.cie.2018.09.052

41.

Nabil

D. H.

Al Amin

Baldacci

(2024). Enhancing resilience in transnational E-commerce supply chains: Critical factors, perspectives and strategic action plan. Heliyon, 10(10), e31274. https://doi.org/10.1016/j.heliyon.2024.e31274

42.

Nakandala

Chen

Chikweche

(2025). SME Supply chain resilience in disruptive times: The effects of supply chain robustness, access to government assistance and disruption intensity. Business Process Management Journal, 31(2), 467–496. https://doi.org/10.1108/BPMJ-02-2024-0073

43.

Ortíz-Barrios

M. A.

Madrid-Sierra

S. L.

Petrillo

Quezada

L. E.

(2025). A novel approach integrating IF-AHP, IF-DEMATEL and CoCoSo methods for sustainability management in food digital manufacturing supply chain systems. Journal of Enterprise Information Management, 38(4), 1281–1326. https://doi.org/10.1108/JEIM-04-2023-0199

44.

Özdemirci

Yüksel

Dinçer

Eti

(2023). An assessment of alternative social banking systems using T-spherical fuzzy TOP-DEMATEL approach. Decision Analytics Journal, 6, 100184. https://doi.org/10.1016/j.dajour.2023.100184

45.

Yan

(2024). Geographical relevance-based multi-period optimization for e-commerce supply chain resilience strategies under disruption risks. International Journal of Production Research, 62(7), 2455–2482. https://doi.org/10.1080/00207543.2023.2217937

46.

Wang

Zhang

Wang

(2023). Developing supply chain resilience through integration: An empirical study on an e-commerce platform. Journal of Operations Management, 69(3), 477–496. https://doi.org/10.1002/joom.1226

47.

Rehman

O. U.

Ali

(2022). Enhancing healthcare supply chain resilience: Decision-making in a fuzzy environment. The International Journal of Logistics Management, 33(2), 520–546. https://doi.org/10.1108/IJLM-01-2021-0004

48.

Rezaei

Wang

Tavasszy

(2015). Linking supplier development to supplier segmentation using Best Worst Method. Expert Systems with Applications, 42(23), 9152–9164. https://doi.org/10.1016/j.eswa.2015.07.073

49.

Russo

Masorgo

Gligor

D. M.

(2022). Examining the impact of service recovery resilience in the context of product replacement: The roles of perceived procedural and interactional justice. International Journal of Physical Distribution & Logistics Management, 52(8), 638–672. https://doi.org/10.1108/IJPDLM-07-2021-0301

50.

Safavi-Naini

Sheppard

N. P.

Uehara

(2004). Import/export in digital rights management. In: Fourth ACM Workshop on Digital Rights Management (DRM 2004), 99–110. https://doi.org/10.1145/1029146.1029163

51.

Sawik

(2013). Selection of resilient supply portfolio under disruption risks. Omega, 41(2), 259–269. https://doi.org/10.1016/j.omega.2012.05.003

52.

Shan

Bai

Fan

Shi

Yang

(2023). Enabling roles of integration and resilience for sustainable supply chain performance: An empirical study on China's e-commerce platforms. Applied Economics, 55(60), 7079–7093. https://doi.org/10.1080/00036846.2023.2186354

53.

Suali

A. S.

Srai

J. S.

Tsolakis

(2024). The role of digital platforms in e-commerce food supply chain resilience under exogenous disruptions. Supply Chain Management: An International Journal, 29(3), 573–601. https://doi.org/10.1108/SCM-02-2023-0064

54.

Torres Vergara

J. I.

Saucedo Martínez

J. A.

Olivo Lucio

(2024). Resilient and sustainable supply chain criteria for performance evaluation: Selection and ranking through fuzzy Delphi. Benchmarking: An International Journal, 31(3), 799–823. https://doi.org/10.1108/BIJ-05-2022-0283

55.

Tukamuhabwa

B. R.

Stevenson

Busby

Zorzini

(2015). Supply chain resilience: Definition, review and theoretical foundations for further study. International Journal of Production Research, 53(18), 5592–5623. https://doi.org/10.1080/00207543.2015.1037934

56.

Tuş

Aytaç Adalı

(2019). The new combination with CRITIC and WASPAS methods for the time and attendance software selection problem. OPSEARCH, 56(2), 528–538. https://doi.org/10.1007/s12597-019-00371-6

57.

Wang

Fang

(2020). Research on E-commerce supply chain design based on MVC model and virtual image technology. IEEE Access, 8, 98295–98304. https://doi.org/10.1109/ACCESS.2020.2996675

58.

Wang

Yan

Jia

Chen

(2023). Building supply chain resilience through ambidexterity: An information processing perspective. International Journal of Logistics Research and Applications, 26(2), 172–189. https://doi.org/10.1080/13675567.2021.1944070

59.

Xuan

H. A.

Trinh

V. V.

Techato

Phoungthong

(2022). Use of hybrid MCDM methods for site location of solar-powered hydrogen production plants in Uzbekistan. Sustain Energy Technol Assess, 52, 101979. https://doi.org/10.1016/j.seta.2022.101979

60.

Yager

R. R.

(2014). Pythagorean membership grades in multicriteria decision making. IEEE Transactions on Fuzzy Systems, 22(4), 958–965. https://doi.org/10.1109/TFUZZ.2013.2278989

61.

Yager

R. R.

(2017). Generalized orthopair fuzzy sets. IEEE Transactions on Fuzzy Systems, 25(5), 1222–1230. https://doi.org/10.1109/TFUZZ.2016.2604005

62.

Yan

Chen

Jia

Liu

(2023). Building supply chain resilience through collaborative innovation: An information processing theory perspective. Journal of Digital Economy, 2, 233–243. https://doi.org/10.1016/j.jdec.2024.01.004

63.

Zadeh

L. A.

(1965). Fuzzy sets. Information and Control, 8(3), 338–353. https://doi.org/10.1016/S0019-9958(65)90241-X

64.

Zadeh

L. A.

(1972). A fuzzy-set-theoretic interpretation of linguistic hedges. Journal of Cybernetics, 2(3), 4–34. https://doi.org/10.1080/01969727208542910

65.

Zaman

S. I.

Khan

S. A.

Qabool

Gupta

(2023). How digitalization in banking improve service supply chain resilience of e-commerce sector? A technological adoption model approach. Operations Management Research, 16(2), 904–930. https://doi.org/10.1007/s12063-022-00341-0

66.

Zhang

Jia

You

J. X.

(2023). Striking a balance between supply chain resilience and supply chain vulnerability in the cross-border e-commerce supply chain. International Journal of Logistics Research and Applications, 26(3), 320–344. https://doi.org/10.1080/13675567.2021.1948978

67.

Zhou

X. Y.

Yan

Khu

S.-T.

Yang

Zhao

(2023). Influence of Russia-Ukraine War on the global energy and food security. Resources, Conservation and Recycling, 188, 106657. https://doi.org/10.1016/j.resconrec.2022.106657

A Multi-Stage T-SF MCDM Framework for Assessing Vulnerabilities in Cross-Border Logistics Chains

Abstract

Keywords

1 Introduction

2.1 Literature Review Procedure Based on Web of Science and Scopus

2.4 Overview of Fuzzy-MCDM Techniques and Applications in SCRE

2.5 Overview of Fuzzy-MCDM Techniques and Applications in CBECSC Resilience

3 Methodology

3.1 Overview of Fuzzy Set Theories

3.1.1 Ordinary Fuzzy Sets (FSs)

Table 6. Linguistic Scale for T-SF DEMATEL Method. Scale ( m , n , p ) Score Index (SI) No Influence (NI) (0, 0.3, 0.2) 0 Weak Influence (WI) (0.35, 0.25, 0.25) 1 Moderate Influence (MI) (0.6, 0.2, 0.35) 2 Strong Influence (SI) (0.85, 0.15, 0.45) 3

4.1 Experts Panel

Table 12. The T-SF COCOSO Results. Stra. k a i k b i k c i k i Rank A1 0.1657 3.0135 0.9811 2.1750 5 A2 0.1673 3.6928 0.9908 2.4660 3 A3 0.1689 4.9697 1.0000 2.9894 1 A4 0.1688 4.8586 0.9997 2.9450 2 A5 0.1625 2.0000 0.9625 1.7206 6 A6 0.1668 3.5814 0.9881 2.4177 4

5.1 Discussion

5.2 Comparative Analysis

Table 13. T-SF TOPSIS Results. Strategies D( X i , X − ) D( X i , X * ) Closeness Coefficient Rank A1 0.066 0.061 0.5201 5 A2 0.069 0.059 0.5407 3 A3 0.074 0.058 0.5629 1 A4 0.071 0.060 0.5426 2 A5 0.056 0.079 0.415 6 A6 0.066 0.060 0.5247 4

6.1 Conclusion

6.2 Theoretical Implications

6.3 Practical Implications

6.4 Limitations and Future Work

Footnotes

Acknowledgements

ORCID iDs

Ethical Approval

Author Contributions

Funding

Declaration of Conflicting Interests

Data Availability Statement

References

Table 6.
Linguistic Scale for T-SF DEMATEL Method.

Scale $(m, n, p)$ Score Index (SI)

No Influence (NI) (0, 0.3, 0.2) 0

Weak Influence (WI) (0.35, 0.25, 0.25) 1

Moderate Influence (MI) (0.6, 0.2, 0.35) 2

Strong Influence (SI) (0.85, 0.15, 0.45) 3

Table 12.
The T-SF COCOSO Results.

Stra. $k a_{i}$ $k b_{i}$ $k c_{i}$ $k_{i}$ Rank

A1 0.1657 3.0135 0.9811 2.1750 5

A2 0.1673 3.6928 0.9908 2.4660 3

A3 0.1689 4.9697 1.0000 2.9894 1

A4 0.1688 4.8586 0.9997 2.9450 2

A5 0.1625 2.0000 0.9625 1.7206 6

A6 0.1668 3.5814 0.9881 2.4177 4

Table 13.
T-SF TOPSIS Results.

Strategies D( $X_{i}$ , $X^{-}$ ) D( $X_{i}$ , $X *$ ) Closeness Coefficient Rank

A1 0.066 0.061 0.5201 5

A2 0.069 0.059 0.5407 3

A3 0.074 0.058 0.5629 1

A4 0.071 0.060 0.5426 2

A5 0.056 0.079 0.415 6

A6 0.066 0.060 0.5247 4