Systems

34 pages, 874 KB

Open AccessArticle

Same Participation Count, Different Memorization: A Controlled Canary Study of Participation Timing in Federated LoRA Fine-Tuning

by Jongwook Kim

Systems 2026, 14(10), 1319; https://doi.org/10.3390/systems14101319 (registering DOI) - 17 Oct 2026

Abstract

Federated low-rank adaptation (LoRA) enables multiple clients to fine-tune a shared language model without centralizing private text. Under partial participation, however, clients with the same participation count may contribute at very different stages of training, and participation count alone does not capture this [...] Read more.

Federated low-rank adaptation (LoRA) enables multiple clients to fine-tune a shared language model without centralizing private text. Under partial participation, however, clients with the same participation count may contribute at very different stages of training, and participation count alone does not capture this temporal structure. It remains unclear whether equally participating clients face different memorization risks depending on when and how their contributions occur. To study this question, controlled paired schedules are constructed that vary the temporal location and within-window pattern of a target client’s participation while holding its participation count, corpus, local training budget, and canaries fixed. Using federated instruction tuning on a cybersecurity corpus, this study evaluates final-model memorization through candidate pool exposure and within-pool top-K rates, then analyzes its evolution using round-wise secret loss trajectories. The temporal location effect is further tested across model families and LoRA aggregation settings, and its mitigation is examined using record-level differentially private stochastic gradient descent (DP-SGD). The results show that later participation substantially increases final-model memorization despite the same participation count, while the exploratory within-window comparison shows a smaller difference. In particular, in the main Qwen experiment, later participation increased mean final candidate pool exposure by 3.684 bits. Temporal diagnostics indicate that late schedules leave fewer opportunities for subsequent weakening before release and that immediate local writing varies across participation stages. The temporal location effect was also observed across the tested model and LoRA settings. Across the four evaluated DP-SGD settings, record-level DP-SGD substantially reduced candidate pool exposure and the systematic late–early advantage observed under clipping alone was not observed, with a corresponding reduction in predictive utility. Overall, these results show that in the tested controlled setting, release-relative participation timing can materially affect synthetic canary memorization even when participation counts are equal. Full article

(This article belongs to the Special Issue Next-Generation Cybersecurity Intelligence: AI-Driven Defense, Quantum-Safe Systems, and Resilient Digital and Environmental Ecosystems)

22 pages, 1489 KB

Open AccessArticle

Insurance Claim Screening Under Capacity Constraints: Model Choice, Review Requirements, and Economic Value

by Kiwoong Byun and Injun Hwang

Systems 2026, 14(10), 1318; https://doi.org/10.3390/systems14101318 (registering DOI) - 17 Oct 2026

Abstract

Insurers must decide which personal insurance claims to investigate and how much review to undertake with limited resources. Using linked insurance and personal credit records covering 2,467,481 claims in South Korea, we provide large-scale empirical evidence on prioritizing claims for investigation. We compare [...] Read more.

Insurers must decide which personal insurance claims to investigate and how much review to undertake with limited resources. Using linked insurance and personal credit records covering 2,467,481 claims in South Korea, we provide large-scale empirical evidence on prioritizing claims for investigation. We compare Logistic Regression, Random Forest, and CatBoost for claim screening, with the joint outcome of selection for investigation and subsequent claim denial as the detection target. We evaluate the capture of denied claims at common review volumes, the review requirements for common recall targets, and potential economic value. To capture 90% of denials, CatBoost required 42.9% fewer expected reviews than Logistic Regression in the test sample. The economic analysis further shows that the review volume maximizing the estimated net value depends on the assumed investigation cost and that additional detection does not necessarily increase net value. Random Forest and CatBoost produce relatively similar maximum scenario value when each model is evaluated at its own value-maximizing review volume. These findings highlight the importance of considering screening model choice and investigation resource allocation together. Full article

(This article belongs to the Special Issue Complex Financial Systems: Dynamics, Risk, and Resilience)

35 pages, 679 KB

Open AccessArticle

Capacity Investment Under Tariff Risk: Rigid Export, Localization, and Flexibility

by Xiaowei Li, Juliang Zhang and Yue Sun

Systems 2026, 14(10), 1317; https://doi.org/10.3390/systems14101317 (registering DOI) - 17 Oct 2026

Abstract

This study examines firms’ capacity-investment choices among export-oriented rigid capacity, localized capacity, and flexible capacity under tariff risk. We develop a single-firm, two-region, two-stage stochastic optimization model with recourse in which capacity is committed before the tariff state is realized, while production and [...] Read more.

This study examines firms’ capacity-investment choices among export-oriented rigid capacity, localized capacity, and flexible capacity under tariff risk. We develop a single-firm, two-region, two-stage stochastic optimization model with recourse in which capacity is committed before the tariff state is realized, while production and flexible-capacity allocation are determined afterward. We derive analytical optimal decisions for each mode and establish explicit pairwise dominance conditions. The results identify three distinct responses to tariff risk. Export-oriented rigid capacity preserves the cost advantage of production in the lower-cost region but can generate tariff-induced underutilization when the high-tariff state occurs. Localization eliminates cross-border tariff exposure at the cost of higher local production and investment costs. Flexible capacity instead mitigates capacity mismatch through state-contingent cross-market reallocation. Its value is conditional: when the reallocation benefit is insufficient to offset the flexibility premium, rigid export strictly dominates; otherwise, partial or complete reallocation can make flexibility attractive. Model-based numerical analyses quantify capacity utilization and reallocation and characterize the regions in which each mode is optimal. Robustness analyses further show that inexpensive post-realization capacity expansion can partially substitute for ex ante flexibility, while a common rigid-export–flexibility–localization switching pattern remains highly stable under moderate joint parameter perturbations but can change under wider perturbations. Overall, optimal capacity investment reflects the interaction among production-cost efficiency, tariff avoidance, tariff-induced underutilization, and state-contingent capacity reallocation. Full article

(This article belongs to the Section Supply Chain Management)

41 pages, 1255 KB

Open AccessArticle

A Data-Driven Decision-Support Framework for Municipal Road-Watering Operations: Environmental Priority Scoring and Route Optimization

by Gabeen Kim, Seoyi Kim, Youjin Kwak, Ian Park and Jaeseok Huh

Systems 2026, 14(10), 1316; https://doi.org/10.3390/systems14101316 (registering DOI) - 17 Oct 2026

Abstract

Dust lifted from road surfaces by passing vehicles is a recurring source of urban particulate-matter exposure. Municipal road-watering operations must allocate limited vehicle time and water capacity across road segments with different service priorities while accounting for refilling, accessibility, and repeated-service requirements. This [...] Read more.

Dust lifted from road surfaces by passing vehicles is a recurring source of urban particulate-matter exposure. Municipal road-watering operations must allocate limited vehicle time and water capacity across road segments with different service priorities while accounting for refilling, accessibility, and repeated-service requirements. This study proposes an integrated decision-support framework for daily road-watering and applies it to Dongjak-gu, Seoul, Republic of Korea. The framework combines data-driven urgency scoring based on floating population, traffic volume, and PM₁₀/PM_2.5 hazard-day frequencies with a selective multi-vehicle, multi-trip arc-routing model and heuristic solution procedures. Its distinguishing formulation feature is a completion-conditioned reward: a road segment contributes its priority score only when all required watering passes are completed within the planning horizon. The case study includes 2213 candidate road segments covering approximately 208 km. Over 30 runs, the proposed genetic algorithm (GA) improves the priority-weighted objective by 35.24% relative to the Phase I constructive solution (mean

Z = 127.13

, 95% CI [125.88, 128.39]). The practice-inspired route-corridor baseline attains a higher aggregate objective through broader contiguous coverage, whereas the GA concentrates service more strongly on higher-urgency roads, completing 67.3% of C4 segments compared with 21.6% under the baseline. Under the same decoder-call budget, adaptive large neighborhood search achieves a higher mean objective value than the GA (130.16 vs. 127.13). These results position the framework as a decision-support architecture for comparing feasible service-allocation strategies rather than as a claim of GA dominance. Full article

(This article belongs to the Special Issue Modeling and Simulation for Decision Support in Urban and Transportation Systems)

26 pages, 8087 KB

Open AccessArticle

Extended Reality Research for Cultural Heritage in the European Union: Insights from a Bibliometric Analysis

by Junwen Lu, Guillermo García-Badell and Joan B. Rodriguez

Systems 2026, 14(10), 1315; https://doi.org/10.3390/systems14101315 (registering DOI) - 17 Oct 2026

Abstract

This study conducts a bibliometric analysis of research on extended reality (XR) technologies in cultural heritage research in the European Union (EU). It maps the overall research landscape and offers insights to inform the future development of XR in this field. Despite the [...] Read more.

This study conducts a bibliometric analysis of research on extended reality (XR) technologies in cultural heritage research in the European Union (EU). It maps the overall research landscape and offers insights to inform the future development of XR in this field. Despite the growing interest in XR within the heritage sector, existing reviews tend to focus on specific technologies or application areas rather than providing a comprehensive overview, particularly in the European context. Europe remains a leading region in cultural heritage practice and research. To address this, the study analyses 1368 publications indexed in the Web of Science Core Collection between 1989 and 2025, affiliated with the 27 EU member states. The analysis uses Bibliometrix, VOSviewer and CiteSpace to examine publication trends, geographical distribution, collaboration networks, keyword patterns and thematic evolution. The results reveal that the annual output increased by an average of 11.35%, accelerating after 2012 and peaking at 174 publications in 2022. Italy, Spain, and Greece account for almost 60% of publications, with Italy contributing 36.2% alone, whereas smaller countries such as Denmark, Belgium, and the Netherlands demonstrate the highest rates of international co-authorship. Research hotspots have shifted from technology-driven 3D documentation and single-technology experiments towards integrated XR, digital twins, HBIM, and user-centred immersive experiences. These findings synthesise a structural framework comprising three interconnected dimensions that shape the field’s knowledge landscape. The first is a value-oriented dimension addressing key heritage aims, including education, virtual archaeology, museum engagement, and architectural heritage conservation. The second is a technological dimension covering core foundations such as virtual reality, augmented reality, mixed reality, 3D modelling, and photogrammetry. The third is a digital implementation dimension focusing on practical approaches such as immersive experiences, storytelling, gamification, and interactivity. The study also identifies three priorities for future research: developing comprehensive technical guidelines, investigating broader socio-technical influences, and strengthening cross-cultural comparative studies. Full article

► Show Figures

Figure 1

20 pages, 645 KB

Open AccessArticle

Digital Capability Configurations and ESG Performance in International Megaprojects: An fsQCA Approach

by Nannan Wang, Huaqin Li and Zhankun Liu

Systems 2026, 14(10), 1314; https://doi.org/10.3390/systems14101314 (registering DOI) - 17 Oct 2026

Abstract

Environmental, social, and governance (ESG) performance is vital for international megaprojects, but current firm-level research ignores the role of digital capabilities in project delivery. Based on dynamic capability theory, this paper examines how the combinations of digital capabilities (sensing, analysis, connecting, integration, and [...] Read more.

Environmental, social, and governance (ESG) performance is vital for international megaprojects, but current firm-level research ignores the role of digital capabilities in project delivery. Based on dynamic capability theory, this paper examines how the combinations of digital capabilities (sensing, analysis, connecting, integration, and agility) jointly shape the project-level ESG performance of megaprojects. By combining fuzzy-set qualitative comparative analysis (fsQCA) and case study methods and using multi-source text data from 35 Belt and Road Initiative (BRI) megaprojects, an evaluation framework is developed to investigate effective causal paths. The results show that no single digital capability constitutes a necessary condition for high ESG performance. Instead, high performance is achieved through equifinality driven by non-redundant capability combinations. Specifically, analysis-oriented configurations support environmental performance through evidence-based decision-making. Sensing-oriented configurations are the basis of social performance to identify stakeholder risks. Connecting- and agility-oriented configurations improve governance performance by helping cross-organizational coordination. Suggestions are provided for practitioners, including aligning digital investment with specific ESG priorities, developing complementary capability configurations that match project governance priorities, and strengthening digital information sharing and adaptive response. This study helps researchers and practitioners understand the configurational effects of digital capabilities on the ESG of international megaprojects. Full article

(This article belongs to the Section Artificial Intelligence and Digital Systems Engineering)

► Show Figures

Figure 1

20 pages, 4660 KB

Open AccessArticle

Identifying the Achilles’ Heel: Using Complex Network Analysis to Pinpoint Systemic Vulnerabilities in the Global Food Supply Network

by Xin Dai, Xiaoyu Jiang, Rui Han, Zhong Li, Yongchao Ding and Ye Liu

Systems 2026, 14(10), 1313; https://doi.org/10.3390/systems14101313 - 16 Oct 2026

Abstract

Systemic risk in the global staple food trade network has become a critical threat to global food security amid escalating geopolitical frictions and frequent extreme weather events. This study constructs directed weighted trade networks for wheat, rice, maize and soybean covering 196 economies, [...] Read more.

Systemic risk in the global staple food trade network has become a critical threat to global food security amid escalating geopolitical frictions and frequent extreme weather events. This study constructs directed weighted trade networks for wheat, rice, maize and soybean covering 196 economies, and applies a cascading failure model to compare cross-crop structural vulnerability. Simulation results reveal stark heterogeneous risk profiles: a shock originating from a top rice exporter in India spreads across two cascading rounds and impacts 25.5% of total network nodes, while a parallel shock from a top soybean exporter in Brazil triggers nine propagation rounds affecting 86.2% of nodes; by comparison, equivalent shocks to wheat and maize networks only affect 22.4% and 24.5% of nodes respectively. Under identical targeted attack conditions, rice and soybean networks suffer sharp connectivity deterioration after removing a small number of core exporter nodes, while wheat and maize networks retain relatively higher residual connectivity under the same attack intensity, reflecting that crop trade systems with higher supply concentration exhibit greater systemic vulnerability. This work adopts a single-year static network framework; future research will extend to multi-year dynamic network tracking and incorporate adaptive supply-chain feedback mechanisms. Despite this boundary, the study fills the research gap in cross-crop systemic risk comparison, providing robust empirical evidence for differentiated global food supply-chain resilience governance. Full article

24 pages, 463 KB

Open AccessArticle

Negotiating Completion in Agentic Software Work: Acceptance Boundaries and Deferred Requirements

by Minyang Ling, Ju Zou and Xintong Liu

Systems 2026, 14(10), 1312; https://doi.org/10.3390/systems14101312 - 16 Oct 2026

Abstract

Coding agents increasingly participate in software development, but submitting an implementation does not settle whether work is complete. This qualitative process comparison examines 30 diagnostic pull requests from 22 GitHub repositories within a frame of 1782 contributions from 120 repositories. Manual screening found [...] Read more.

Coding agents increasingly participate in software development, but submitting an implementation does not settle whether work is complete. This qualitative process comparison examines 30 diagnostic pull requests from 22 GitHub repositories within a frame of 1782 contributions from 120 repositories. Manual screening found substantive contestation in 516 frame contributions (29.0%). The analysis draws on public discussions, reviews, commits, tests, and linked artifacts to trace how participants address unmet or unproven requirements. Using acceptance boundary as a sensitizing distinction, the analysis connects conditions stated in review to their enactment in evaluation artifacts and subsequent revision. Participants may retain requirements for repair, delimit a smaller acceptable increment, pursue an alternative implementation, or terminate an approach. Tests enact these decisions through repair, authorized exclusion, and restoration. In sqruff, issue closure accompanied partial repair and renewed deferral; a subsequent rollback reversed the accepted changes. Integration, issue closure, and requirement satisfaction therefore represent distinct accomplishments. The process model identifies conditional, overlapping paths connecting local acceptance to continuing work. It explains collaborative completion in agent-associated software development and shows how documented interface constraints and human intervention shape subsequent work. Full article

(This article belongs to the Special Issue Artificial Intelligence in Socio-Technical Systems)

► Show Figures

Figure 1

28 pages, 7747 KB

Open AccessArticle

Conceptualising Digital Transformation Through Recursive Capability Constriction: Evidence from Australian NDIS Providers

by Leela Waheed, Shammi Aktar and Aminath Waseela

Systems 2026, 14(10), 1311; https://doi.org/10.3390/systems14101311 - 16 Oct 2026

Abstract

Digital transformation is a strategic priority across care-intensive service sectors, yet many organisations struggle to realise its intended benefits. Existing research often examines technological, organisational, and environmental barriers separately, offering limited insight into their interconnections within organisational systems. Drawing on Systems Theory and [...] Read more.

Digital transformation is a strategic priority across care-intensive service sectors, yet many organisations struggle to realise its intended benefits. Existing research often examines technological, organisational, and environmental barriers separately, offering limited insight into their interconnections within organisational systems. Drawing on Systems Theory and Sociotechnical Systems (STS) Theory, this study investigates digital transformation among small and medium providers within Australia’s National Disability Insurance Scheme (NDIS). Using an interpretive qualitative design, semi-structured interviews were conducted with 15 participants from 15 distinct NDIS provider organisations in Sydney, Australia, followed by reflexive thematic analysis and systems-oriented interpretation. Participants described recurring associations among technological misalignment, fragmented software, adaptive workarounds, increased administrative demands, and constraints on organisational capacity for digital improvement, with regulatory and resource pressures interacting with these technological and organisational conditions. Building on these cross-sectional patterns, this study develops Recursive Capability Constriction (RCC), a systems-based conceptual framework that proposes how technological misalignment, compensatory adaptation, administrative expansion, and constrained organisational capability for digital improvement may become interconnected through reinforcing relationships. Rather than representing an inevitable or empirically established causal sequence, RCC provides a conditional explanation of how local adaptations may become associated with constraints on subsequent digital improvement. By shifting attention from isolated implementation barriers to interconnected sociotechnical relationships, RCC extends digital transformation research and provides practical guidance for managing digital change in regulated care environments. Full article

(This article belongs to the Section Systems Theory and Methodology)

► Show Figures

Figure 1

26 pages, 1048 KB

Open AccessArticle

Augmenting Process Failure Mode and Effects Analysis with Fuzzy Cognitive Mapping for Dynamic Manufacturing Risk Assessment

by Amin Sehati and Steven A. Gray

Systems 2026, 14(10), 1310; https://doi.org/10.3390/systems14101310 - 16 Oct 2026

Abstract

Conventional Process Failure Mode and Effects Analysis (PFMEA) is linear and does not explicitly include a systems-based perspective that accounts for the complexity in which all manufacturing systems operate. In an illustrative case study using hypothetical inputs, this study augments traditional PFMEA methods [...] Read more.

Conventional Process Failure Mode and Effects Analysis (PFMEA) is linear and does not explicitly include a systems-based perspective that accounts for the complexity in which all manufacturing systems operate. In an illustrative case study using hypothetical inputs, this study augments traditional PFMEA methods with a Fuzzy Cognitive Mapping (FCM) approach. We suggest a Leverage Impact Number (LIN) that quantifies modeled system-wide risk reduction to include the impacts of individual intervention options available to decision-makers to better quantify risk, but also the relative contribution of management intervention to reduce those risks. Rank comparisons examine changes in priorities and an analysis of concepts covered only by the FCM considers organizational factors. Monte Carlo perturbations assess sensitivity to causal weights, and one-way analyses examine risk weighting, mitigation levels, driver baselines, transfer-function steepness, and the update rule. When traditional PFMEA is compared with our FCM-augmented approach, our results indicate that LIN and Risk Priority Number (RPN) rankings are moderately correlated in this illustrative case, with the augmented approach retaining key PFMEA priorities while extending the analysis to modeled system-wide intervention effects. Under reference assumptions, combined stressors increase modeled risk by approximately the sum of their separate effects. These results demonstrate additional decision support information under the same stated assumptions when using PFMEA and our FCM-augmented method. Full article

(This article belongs to the Section Systems Engineering)

► Show Figures

Figure 1

52 pages, 1547 KB

Open AccessArticle

One Step Ahead of the Crisis: Evaluating Digital Solutions for Improving Humanitarian Logistics Based on Literature-Informed Insights Relevant to Humanitarian Organizations

by Aleksa Maravić, Mladen Krstić, Snežana Tadić and Miloš Veljović

Systems 2026, 14(10), 1309; https://doi.org/10.3390/systems14101309 - 16 Oct 2026

Abstract

Digitalization represents an important potential for improving humanitarian logistics (HL) and supporting humanitarian organizations (HOs) in shifting from a reactive to a proactive approach, i.e., acting “one step ahead of the crisis”. The aim of this study is to identify the most suitable [...] Read more.

Digitalization represents an important potential for improving humanitarian logistics (HL) and supporting humanitarian organizations (HOs) in shifting from a reactive to a proactive approach, i.e., acting “one step ahead of the crisis”. The aim of this study is to identify the most suitable digital solution for improving HL by considering it as a complex adaptive socio-technical system and applying a new hybrid Multi-Criteria Decision-Making (MCDM) model. The model integrates the CILOS (Criterion Impact Loss) method for determining the objective weights of the criteria and the COBRA (Comprehensive Distance-Based Ranking) method for evaluating and ranking the alternatives. Four digital technologies were analyzed—Internet of Things (IoT), Big Data technologies, Enterprise Resource Planning (ERP) systems, and mobile technologies—based on six criteria: preparedness and resilience, logistics efficiency, integration and adaptability, economic feasibility, reliability and information security, and social acceptability. The results indicate that the Internet of Things represents the most suitable solution for HOs, followed by ERP systems, mobile technologies, and Big Data technologies. The main contributions of the study include establishing a new systemic framework for evaluating digital technologies in HL that supports a shift from a reactive to a proactive approach (“one step ahead of the crisis”) under conditions of crisis-related uncertainty, as well as introducing a hybrid CILOS-COBRA Multi-Criteria Decision-Making (MCDM) model for evaluating digital solutions in HL. The selection of a digital solution should be based on its contribution to the operational capabilities of HOs, particularly in terms of preparedness, efficiency, reliability, and resilience, thereby supporting a “one step ahead of the crisis” approach. Full article

(This article belongs to the Section Systems Engineering)

► Show Figures

Figure 1

38 pages, 1594 KB

Open AccessArticle

How Does Downstream Customer SRDI “Little Giant” Certification Drive Supply Chain Diversification Among Upstream Suppliers? Evidence from China

by Xuan Ling and Lin Song

Systems 2026, 14(10), 1308; https://doi.org/10.3390/systems14101308 - 16 Oct 2026

Abstract

Growing uncertainty in the external environment has heightened concerns about supply chain security and resilience. Existing studies mainly examine the effects of China’s Specialized, Refined, Distinctive, and Innovative (SRDI) “Little Giant” certification on certified firms themselves, while its implications for the supply chain [...] Read more.

Growing uncertainty in the external environment has heightened concerns about supply chain security and resilience. Existing studies mainly examine the effects of China’s Specialized, Refined, Distinctive, and Innovative (SRDI) “Little Giant” certification on certified firms themselves, while its implications for the supply chain configuration of connected firms remain less understood. Using the staggered timing of certification, this study uses customer–supplier relationship data from Chinese listed firms over the period 2012–2025 and employs a staggered difference-in-differences model to examine whether downstream customer certification is associated with changes in supply chain diversification among upstream suppliers. The results show that “Little Giant” certification significantly reduces the concentration of upstream suppliers’ major trading relationships, leading to a more dispersed distribution of transaction shares across major partners. This finding remains stable across a range of alternative specifications, while an interaction-weighted event-study analysis provides complementary evidence on heterogeneous treatment timing and dynamic patterns. The mechanism analysis provides suggestive evidence consistent with a reputation-related channel and operating-resource adjustment: certified customers are associated with greater positive media exposure for their suppliers and higher operating working capital investment, while changes in current assets provide complementary evidence of operating-resource adjustment. Further analysis indicates that the estimated association is more evident in procurement relationships than in sales relationships, although the difference between the two estimates is not statistically significant. The estimated association also tends to be stronger among suppliers that exhibited overinvestment before certification and those with larger pre-certification market shares. These findings extend research on the economic consequences of SRDI certification from certified firms to their supply chain partners and suggest that firm-targeted industrial policy may be associated with cross-firm adjustments in major trading relationships through existing customer–supplier linkages. Full article

(This article belongs to the Special Issue Innovation and Sustainability Management: Challenges for Small- and Medium-Sized Enterprises and Their Supply Chains)

► Show Figures

Figure 1

25 pages, 1135 KB

Open AccessArticle

The Role of Supply Chain Innovation Support Policies as a Determinant of the Interdependence Between Digitalization and Greening

by Weijia Chen, Tao Wang and Meigen Huang

Systems 2026, 14(10), 1307; https://doi.org/10.3390/systems14101307 - 15 Oct 2026

Abstract

This study examines the association between supply chain innovation support policies and the dynamic interdependence between firms’ digitalization and greening. Using a panel of 320 A-share listed firms from 11 provinces in the Yangtze River Basin over 2010–2024, we construct a dynamic measure [...] Read more.

This study examines the association between supply chain innovation support policies and the dynamic interdependence between firms’ digitalization and greening. Using a panel of 320 A-share listed firms from 11 provinces in the Yangtze River Basin over 2010–2024, we construct a dynamic measure of digital–green interdependence using the asymmetric dynamic conditional correlation (ADCC) framework. The resulting dynamic correlation measure is then examined in relation to supply chain innovation support policies and firm-level characteristics. The baseline results show that supply chain innovation support policies are positively associated with digital–green interdependence. Subsample analysis indicates that this association is stronger during the post-policy period. Quantile regression further reveals distributional heterogeneity, with larger policy coefficients at higher quantiles of digital–green interdependence. Markov-switching analysis identifies distinct regimes in the evolution of digital–green interdependence, with regime transitions occurring around the period of policy implementation. Threshold analysis further indicates a nonlinear relationship between policy support and digital–green interdependence. Robustness analysis based on the Fisher Z-transformed dynamic correlation yields results consistent with the baseline findings. Overall, the findings highlight the dynamic, heterogeneous, and nonlinear characteristics of the relationship between supply chain innovation support policies and firms’ digital–green interdependence. Full article

(This article belongs to the Special Issue Digital Transformation in the Era of Technological Disruption: The Reshaping and Application of Emerging Technologies on Management Models)

► Show Figures

Figure 1

28 pages, 3971 KB

Open AccessArticle

Multi-Level Innovation Ecosystem Analysis of China’s Ice and Snow Equipment Manufacturing: From Firm to Regional to Industrial Perspectives Using fsQCA

by Lanyue Xu and Dingqing Wang

Systems 2026, 14(10), 1306; https://doi.org/10.3390/systems14101306 - 15 Oct 2026

Abstract

This study aims to identify the multi-level configurational pathways through which firm, regional and industrial conditions jointly shape firm innovation performance within China’s ice and snow equipment manufacturing ecosystem. Complex causal interplay among firm capabilities, regional environments and industrial structures shapes innovation ecosystems [...] Read more.

This study aims to identify the multi-level configurational pathways through which firm, regional and industrial conditions jointly shape firm innovation performance within China’s ice and snow equipment manufacturing ecosystem. Complex causal interplay among firm capabilities, regional environments and industrial structures shapes innovation ecosystems in emerging manufacturing sectors, yet variance-based methods that assume symmetric and additive effects while overlooking configurational equifinality continue to dominate the existing literature. The unit of analysis is the individual firm embedded within its regional and industrial context, and the outcome captures firm-level innovation performance as shaped by ecosystem participation. A multi-level configurational framework incorporating firm-level conditions (R&D investment intensity and digital transformation capability), regional-level conditions (policy support intensity and market demand scale) and industrial-level conditions (supply chain completeness and collaborative innovation network density) is built. By combining fuzzy-set qualitative comparative analysis (fsQCA) with necessary condition analysis (NCA), the configurations associated with high firm innovation performance within the ecosystem are identified. This study represents the first application of the combined NCA and fsQCA approach to the ice and snow equipment sector, explicitly operationalizing a two-tier causal logic in which necessary bottleneck conditions constrain the outcome space before sufficient configurations are mapped within that space. Survey and archival data from 86 ice and snow equipment manufacturers across seven provincial-level regions are calibrated and studied. From the NCA results, digital transformation capability and supply chain completeness emerge as individually necessary conditions. The sufficiency analysis uncovers four equifinal configurations leading to high performance. Three distinct configurations linked to the absence of high performance are also obtained from the negation analysis, and causal asymmetry is confirmed. An exploratory cross-level comparison suggests that conditions from all three analytical levels appear as core elements across the four configurations with near-comparable frequency, and the dominant level shifts from one configuration to another rather than being universally fixed. When alternative calibration anchors, frequency thresholds and an enhanced standard analysis are considered, the stability of the solutions is confirmed for three of the four configurations; the fourth (digitally anchored collaboration) is sensitive to the frequency threshold, consistent with its smaller empirical base. These findings enrich innovation ecosystem theory by showing that multi-level conditions interact through complementary and potentially substitutive mechanisms rather than independently, and they give actionable guidance for policymakers and managers who seek to build resilient innovation ecosystems in China’s winter sports equipment sector. Full article

(This article belongs to the Special Issue The Dynamics and Resilience of Innovation Networks in a Disruptive World)

25 pages, 1246 KB

Open AccessArticle

Optimization of Waste Recycling Vehicle Routing in Public Health Emergencies

by Shaoren Wang, Keyi Qian, Jiajia Lin and Linlin Zheng

Systems 2026, 14(10), 1305; https://doi.org/10.3390/systems14101305 - 15 Oct 2026

Abstract

With the frequent occurrence of public health emergencies such as the COVID-19 pandemic, the allocation and coordinated use of emergency supplies around the world have generated large amounts of waste, thus creating an urgent need to improve and optimize emergency reverse logistics systems. [...] Read more.

With the frequent occurrence of public health emergencies such as the COVID-19 pandemic, the allocation and coordinated use of emergency supplies around the world have generated large amounts of waste, thus creating an urgent need to improve and optimize emergency reverse logistics systems. This study proposes a model for optimizing waste recycling vehicle routing to address waste management issues. The model provides ideal transportation solutions for the simultaneous collection of general waste and medical waste. Based on the characteristics of the model, an artificial intelligence algorithm—i.e., an improved genetic algorithm based on bit-string-matching coding and a field restriction operator—is designed and compared with an ant colony optimization algorithm. A synthetic numerical example inspired by the situation in Wuhan, China, is tested using relevant data on the locations of 18 medical facilities mentioned in existing case studies of the COVID-19 pandemic to verify the effectiveness and practicality of the model and the algorithm. This study provides support for decision-making regarding optimization strategies for the reverse logistics of waste generated during public health emergencies. Full article

(This article belongs to the Section Systems Practice in Social Science)

26 pages, 4230 KB

Open AccessArticle

Regional Integration and National New Quality Productive Forces: Evidence from the Beijing–Tianjin–Hebei Region

by Miaomiao Niu and Guohao Li

Systems 2026, 14(10), 1304; https://doi.org/10.3390/systems14101304 - 15 Oct 2026

Abstract

This study integrates a multi-regional input–output model, structural decomposition analysis, a coupling coordination model and gray relational analysis within one framework. It examines the relationship between the Beijing–Tianjin–Hebei (BTH) region’s participation in national economic circulation (NEC) and national new quality productive forces (NQPFs). [...] Read more.

This study integrates a multi-regional input–output model, structural decomposition analysis, a coupling coordination model and gray relational analysis within one framework. It examines the relationship between the Beijing–Tianjin–Hebei (BTH) region’s participation in national economic circulation (NEC) and national new quality productive forces (NQPFs). The results show that China’s NQPF index rose from 0.02 to 0.31 over 2012–2020, with the means of labor making the largest contribution, while growth in green and efficient productivity remained limited. BTH’s participation in NEC exhibited a U-shaped pattern, with the inflection point coinciding with the implementation of the Coordinated Development Strategy in 2015. The consumption effect became the largest contributing component after 2015, whereas the value-added and spillover effects made negative contributions in some periods. Coupling coordination between the NEC embedding system and the national NQPF system shifted from imbalance toward high-quality coordination and strengthened after 2015. Gray relational analysis identified industrial value-added growth and consumption upgrading as the components most strongly associated with the NQPF index, whereas the multiplier and investment effects showed the weakest associations. Full article

(This article belongs to the Section Systems Practice in Social Science)

► Show Figures

Figure 1

32 pages, 369 KB

Open AccessArticle

Naming the Conflict, Choosing the Response: Managerial Framing of Multicultural Conflict as a Systemic Mechanism in Schools

by Chajim Erlanger

Systems 2026, 14(10), 1303; https://doi.org/10.3390/systems14101303 - 15 Oct 2026

Abstract

Educational practitioners encounter conflicts in which the framing of an event is systematically associated with the organizational response that is represented as being available. This study examines that association as it appears in the practitioner’s own account and at the meso level of [...] Read more.

Educational practitioners encounter conflicts in which the framing of an event is systematically associated with the organizational response that is represented as being available. This study examines that association as it appears in the practitioner’s own account and at the meso level of the institution. The corpus consists of 52 case analyses written by serving educational practitioners enrolled in a graduate program in educational systems management. The unit of analysis is the managerial narrative rather than the event itself. Cases were coded for managerial framing and for response type: separation, decisive resolution, and flexible implementation. High agreement was found across three independent blind-coding rounds. Framing as discrimination was associated with decisive resolution in all ten related cases, and framing as difference was associated with separation or flexible implementation in the remaining forty-two cases. A cluster in which writers name a breach of a core norm and draw no conclusion of consequence from it marks the boundary conditions of the pattern, and shows the vocabulary of cultural responsiveness serving as a shield against enforcement rather than as a route to inclusion. The study contributes to a systems-oriented understanding of educational decision making by showing how the categorization of a conflict is associated with the range of responses represented as being available. Full article

(This article belongs to the Special Issue Navigating Educational Leadership Through Systems Approaches)

36 pages, 10385 KB

Open AccessArticle

The Impact of China’s Digital Industrial Cluster Policy on Urban Carbon Quota Development

by Qiutong Xue and Weikai Meng

Systems 2026, 14(10), 1302; https://doi.org/10.3390/systems14101302 - 15 Oct 2026

Abstract

The digital industrial cluster policy is a key policy instrument for fostering the digital economy and facilitating urban low-carbon transition. Given that no official carbon quota allocation system has yet been established at the city level in China, this paper constructs a multidimensional [...] Read more.

The digital industrial cluster policy is a key policy instrument for fostering the digital economy and facilitating urban low-carbon transition. Given that no official carbon quota allocation system has yet been established at the city level in China, this paper constructs a multidimensional urban carbon quota index that integrates social, economic, and ecological dimensions, and uses it to measure the carbon quota level of 297 prefecture-level-and-above cities in China over the period 2006–2024. Treating China’s digital industrial cluster policy as a quasi-natural experiment, we employ a staggered difference-in-differences (DID) design to systematically evaluate the policy’s effect on urban carbon quota and the underlying mechanisms. The results show that ① the digital industrial cluster policy significantly raises the urban carbon quota level; ② mechanism tests indicate that this effect is channeled through three paths—stimulating green technological innovation, promoting industrial structure advancement, and improving electricity energy efficiency; ③ heterogeneity analysis reveals that the effect is more pronounced in western regions, in central cities, and in areas with more advanced digital infrastructure and higher carbon quota endowment; ④ spatial econometric results further show that the policy not only enhances local carbon quota but also generates a significant positive spatial spillover effect, raising the carbon quota level of neighboring cities. These findings offer insights for optimizing urban carbon quota allocation through spatial coordination, path coupling, and targeted policy design. Full article

(This article belongs to the Section Systems Practice in Social Science)

► Show Figures

Figure 1

38 pages, 63651 KB

Open AccessArticle

Stability and Change in China’s Digital Governance Policy: A Bibliometrics-Based Punctuated Equilibrium Analysis Framework

by Chao Yang, Qiming Chen and Yujing Zhou

Systems 2026, 14(10), 1301; https://doi.org/10.3390/systems14101301 - 15 Oct 2026

Abstract

China’s digital governance encompasses both the use of digital technologies in public governance and the governance of digital development. It is a key fulcrum for the systematic reconstruction of China’s governance system and an important measure to enhance governance levels and promote the [...] Read more.

China’s digital governance encompasses both the use of digital technologies in public governance and the governance of digital development. It is a key fulcrum for the systematic reconstruction of China’s governance system and an important measure to enhance governance levels and promote the transformation of government functions. However, systematic analyses of China’s multilevel digital governance policy system remain limited, particularly regarding its evolutionary trajectory, core features and existing gaps. To address these problems, we construct a bibliometrics-based punctuated equilibrium analysis framework to unpack the dynamic progression of China’s digital governance policy system. The analysis covers 6309 policy documents issued between January 2000 and 19 May 2025. We used rule-based natural language processing to identify policy targets and instruments, constructed weighted co-occurrence networks of policy targets, co-issuing agencies, and target–instrument pairs, and calculated eigenvector centrality. We then examined the policy system across four stages and three dimensions: policy targets, policy-issuing agencies, and policy mixes. Finally, we identified changes in the prominence and configuration of targets, institutions, and instruments while also revealing limitations in the existing policy system. The findings provide insights for improving digital governance policies and extend the explanatory scope of punctuated equilibrium theory in a non-Western institutional context. Full article

(This article belongs to the Section Systems Practice in Social Science)

► Show Figures

Figure 1

Journal Description

Systems

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI