We analyze a multi-item production-inventory system, focusing on minimizing the Conditional Value-at-Risk (CVaR) of total inventory holding and backordering costs. Instead of optimizing expected costs, we incorporate risk sensitivity to better align with decision-makers' preferences. The system is modeled as a multi-class M/M/1 queue, where items are produced individually. Due to the complexity of transient queueing behavior and multi-item interactions, we propose five heuristics and evaluate their performance through a numerical experiment with 216 instances. Our results indicate that one heuristic performs particularly well. Additionally, we present ten key observations regarding the impact of economic and technical parameters on heuristic basestock levels and optimality gaps. Notably, the optimal basestock level exhibits significant sensitivity to the number of items but minimal sensitivity to the CVaR risk parameter. We also extend our analysis to scenarios with heterogeneous backorder costs and demand rates across items with additional numerical experiments. Our findings provide practical decision-making tools for managers handling multi-item production-inventory systems with downside risk considerations.
Blockchain technology has revolutionized various industries by offering transparency, security, and decentralization. The critical aspect of blockchain technology is the consensus protocol, which plays a pivotal role in ensuring the integrity and reliability of distributed ledger systems. The selection of an appropriate consensus protocol for a given blockchain application is a complex and multifaceted decision-making process, influenced by various technical, environmental, and operational factors. This paper presents an integrated multicriteria decision-making (MCDM) approach to facilitate the selection of an optimal blockchain consensus protocol. Through a comprehensive evaluation of criteria, including performance, sustainability, incentives, security, and decentralization, our approach provides a robust decision-making framework for consensus protocol assessment. The results prioritize the importance of performance and security factors in blockchain consensus protocol evaluation. The sensitivity analysis is performed to determine the impact of experts' weight coefficients on the result. The results prioritize the importance of performance and security in blockchain consensus protocol selection.
The deterioration rate and selling value of perishable items, such as fruits, milk, vegetables, meat, seafood, and packaged goods, directly impact an organisation's profit. Therefore, to leverage profit, efforts are required to reduce the deterioration rate of these perishable items. However, an excessive amount of freshness-keeping effort may lead to an increase in the selling price, which could result in customer dissatisfaction. This problem has been the subject of investigation by various researchers, who have employed different mathematical models that treat demand as a function of the freshness parameter, along with a constant or variable deterioration rate. However, the freshness parameter and deterioration rate are time-dependent; still, mathematical models that consider both these parameters as functions of time in two-warehouse environments have been developed. Hence, this paper fills this gap by providing mathematical models for two-warehouse inventory systems under First In First Out (FIFO) and Last In First Out (LIFO) dispatching policies. The perishability rate of items is assumed to be Weibull-distributed with respect to time, and demand is a continuous function of the freshness parameter, selling price, and time. Additionally, these mathematical models include parameters such as inflation, price elasticity, and partial backlogging. The primary objective of this study is to maximise profit by optimising the selling price and inventory size. The findings of this study reflect the superiority of the LIFO policy over the FIFO policy under these conditions. Furthermore, the behaviour of these models has been studied through a comprehensive sensitivity analysis, along with an examination of their applicability and managerial implications.
Risk assessment for project portfolio (PP) is pivotal in aligning strategic objectives (SOs) with organizational value propositions. Despite its significance, the impact of project portfolio risks (PPRs) on achieving SOs remains understudied, potentially causing inaccurate assessment. Applying a Bayesian network (BN) to model a tri-layer (PPR-PP-SO) network, this study bridges the gap and forges a direct connection between PPR assessments and SOs. Initially, the PPR criteria are determined, followed by constructing a BN structure to clarify relationships within the three-layer network. To mitigate data scarcity limitations in BN, Spherical Fuzzy Set and Dempster-Shafer theory are introduced to acquire tri-layer BN parameters. Utilizing propagation analysis of the tri-layer BN, the expected impact of risks on SOs is measured to reveal PPR criticality. A numerical example corroborates the model's functionality and feasibility. The results indicate the criticality of one PPR varies substantially under different SOs, implying the importance of pinpointing critical risks specific to each SO.
This research explores the possibility of applying the First Digit Law of Benford (BL1) - a statistical principle stating that in many naturally occurring datasets, the first digit is more likely to be small - in auditing the codification of hospital discharges, using two types of analysis: one involves ten European countries with Organisation for Economic Co-operation and Development (OECD) data plus Romania, grouped by main classes of diseases, and the other applies the technique to the Romanian case through a particular set of 28,000,000 hospitalized events, discharged between 2016 and 2019, grouped by categories. Chi-squared, Sum of Squared Deviances, and Freedman- Watson evaluate the agreement with the Leading Digit Law. Then, our work combines the probability of rejecting Benford Law (BL) and several explanatory variables into a panel binary logit model. The tests applied to the selected OECD countries and Romanian data reveal close conformities by the OECD broad morbidity classes. Particular inquiries applied in the Romanian case show puzzling results. For instance, the disease group with the most years of life lost class and the class with the highest number of discharges fit (BL1). A contrario, the most expensive disease class breaks down the law. Among other interesting results, regression modeling finds that classes with higher cost per episode are more likely to break the BL1, while the share of cases treated in inpatient care for a particular group of diagnostics tends to obey the natural law. We also highlight that anomalies are directly linked with the cost and polarization of cases in university centers and are inversely associated with a possible wider availability of services. Our findings support using this procedure as a primary filter in diagnostic coding audits. The specific set of diagnostics and covariates emphasized here can help the auditor in identifying higher anomaly rates and save time for hospital data mining.
Enterprises increasingly recognize data as a pivotal asset, yet understanding the complex and interdisciplinary nature of data valuation remains a challenge. This paper conducts a systematic literature review, linking data value drivers and approaches to enterprise architecture layers (business, data, application, technology) using The Open Group Architecture Framework (TOGAF) standard. Out of 102 papers, this study identifies seven core data valuation approaches and seven central data value drivers, emphasizing business utility and use case, costs, data security and privacy, and data quality. The findings reveal the pervasive impact of data value across all enterprise architecture layers. The paper consolidates these insights into a conceptual model, establishing a foundation for cross-domain research and practical data valuation solutions for professionals in real-world and academic settings.
Financial market speculative bubbles strongly impact the economy, like the June 2017-February 2018 Bitcoin bubble. Several factors often fuel them, and the role of online communities in influencing such market dynamics is unclear. Understanding the interaction between online discourse and cryptocurrency market behaviour is essential for exploring whether online communities influence speculative market dynamics. While prior research has focused on major platforms like X or Reddit, few studies have systematically analysed blockchain-based platforms' online behaviour in relation to these speculative bubbles. This paper analyses the relationship between the Steemit platform discussion dynamics and the Bitcoin bubble of 2017-2018. Term frequency analysis and correlation methods were used to explore how key terms in Steemit discussions evolved and whether they corresponded with Bitcoin's price changes during the bubble. The contribution of this research is to enhance the understanding of speculative events for risk management and policy decisions by exploring whether there are significant relations among online discussions over market behaviour and investor sentiment. The analysis indicates that while Steemit's discussions largely mirrored the ongoing market sentiment, specific terms exhibited strong positive correlations with Bitcoin price fluctuations, highlighting the role of particular discussions in amplifying speculative sentiment. These results suggest that online discourse may contribute to investor sentiment without necessarily driving market dynamics. This study provides valuable insights into the role of blockchain-based platforms in speculative events, offering implications for risk management and financial policy.
This research introduces an enhanced version of Interpretive Structural Modelling (ISM) within a neutrosophic environment and under plithogenic conditions to address the limitations of conventional ISM. To mitigate these limitations, a novel methodology termed Plithogenic integrated Neutrosophic ISM (P-NISM) is proposed in this research. In P-NISM, neutrosophic sets are utilised to quantify the degree of influence between factors, enabling a more detailed analysis that considers the inherent uncertainties and complexities of decision-making processes. Expert opinions are aggregated using a plithogenic operator, which integrates all sets of opinions, thereby addressing the biases associated with the consensus method. To illustrate the effectiveness of P-NISM, a use case is considered to evaluate the interdependencies among factors influencing the implementation of decarbonisation approaches. Comparative analysis between conventional MICMAC and plithogenic-integrated MICMAC results improves the comprehension of the methodology's utility. This research presents the integration of the opinions of all experts while addressing the drawbacks of the consensus method and the uncertainties in the information.
Marketing analytics and human resource management systems (HRMS) have long been pivotal and cognitive aspects in organizational decision-making, yet the relationship between these two domains needs rather a modest analytical approach that accounts for both departmental dynamics and employee-related factors. The study introduces a novel framework, the Reflection Rotation Equivariant Invariant Quantum Prairie Dog Attention Network (2REI-QuaPDAN), which leverages advanced machine learning techniques to analyze employee data across departments such as Marketing, HR, Sales, IT, and Finance. The dataset, titled "marketing HR analysis," contains 14,999 records with information about various employee metrics, including Marketing Spend, Recruitment Cost, Employee Satisfaction, Attrition Rate, and more. Pre-processing methods, namely Correlation Coefficients and Min-Max Normalization, pre-process the data for further analysis. Feature extraction is carried out using Inception Convolutional Vision Transformers, followed by predictions using the 2REI-QuaPDAN, which integrates reflection-equivariant quantum neural networks with rotation-invariant attention mechanisms. The Model's parameters are optimized with the Prairie Dog Optimization Algorithm (PDOA). The proposed 2REI-QuaPDAN framework achieved an impressive 99.9% prediction accuracy, representing an improvement of over 7-15% compared to existing neural network and analytical models. This significant enhancement demonstrates the Model's superior capability in capturing complex interdependencies between marketing and HR parameters, ensuring more reliable and data-driven organizational insights. This method not only increases the marketing analytics and the human resource management system accuracy, but also provides substantial advantages in terms of scalability and model adaptability, enabling efficient decision-making with high accuracy.
The rapid development of live-streaming e-commerce has challenged traditional dual-channel supply chains, spotlighting the online channel promotion strategies of manufacturers in live-streaming e-commerce environments. This paper explores a supply chain comprising a manufacturer, a traditional retailer, and a live-streaming e-commerce platform. Within this dual-channel framework, the manufacturer has two options for promoting the online direct sales channel: self-promotion or collaboration with streamers through the live-streaming e-commerce platform. In collaborating, the manufacturer can choose to partner with either ordinary streamers or celebrity streamers. The research presents the following findings: First, the promotional effort level of celebrity streamers is the highest in most scenarios. However, interestingly, when the potential market demand rate and sales retention rate of the online channel are high, the manufacturer's self-promotion model exhibits the highest promotional effort level. Second, regardless of the sales retention rate for the online channel, when the spillover effects are weak, and the potential market demand ratio for online is low, the manufacturer maximizes offline sales through self-promotion. Additionally, regardless of changes in the spillover effects, when the sales retention rate and the potential market demand ratio for the online channel are high, the manufacturer achieves the most offline sales using ordinary streamers for promotions. Finally, if the potential market expansion driven by celebrity streamers is substantial and the product retention rate is very high, or both the base commission rate and additional commission rates are low, the manufacturer's optimal strategy is to choose a celebrity streamer for promotions. But as the potential market expansion driven by celebrity streamers decreases and product retention rate gradually declines, or when the streamer's commission rate is high, the manufacturer tends to opt for the self-promotion model.
This study investigates a stochastic production planning problem with a running cost composed of quadratic production costs and inventory-dependent costs. The objective is to minimize the expected cost until production stops when inventory reaches a specified level, subject to a boundary condition. Using probability space and Brownian motion, the Hamilton-Jacobi-Bellman (HJB) equation is derived, and optimal feedback control is obtained. The solution demonstrates desirable monotonicity and convexity properties under specific assumptions. An illustrative example further confirms these results with explicit function properties and a practical application.
We examine the sustainability performance of Smalland Medium-Sized Enterprises (SMEs) across four European countries, focusing on operational, economic, social, and environmental practices. Data were collected through a Likert-scale questionnaire. While Data Envelopment Analysis (DEA) has been widely used to assess SMEs' sustainability, most studies rely on conventional DEA models. This study adopts an interval-scale DEA model to better accommodate interval-scale data. We then apply dimensionality reduction and visualization techniques to explore intra- and inter-country differences at both aggregate and SME levels. Finally, we introduce a novel HHI-GPSM-LASSO framework to assess the causal effect of SMEs' resource allocation strategies on efficiency. This framework integrates a Herfindahl-Hirschman-like index, Generalized Propensity Score Matching, and weighted Lasso regression to uncover the causal relationships between SMEs' resource allocation strategies and efficiencies. While visualization enhances interpretability for practitioners, the causal framework supports strategic and policy decisions. To our knowledge, this is the first DEA study that combines interval-scale DEA, advanced visualization, and causal inference to inform sustainability benchmarking and policy.
Predicting corporate bankruptcy is a critical issue in the financial sector, requiring a comprehensive evaluation of a company's financial health to identify firms that may fail to meet their debt obligations. This issue not only has far-reaching implications for a company's development but also holds significant importance for socio-economic stability. The introduction of clustering algorithms into the field of enterprise risk management represents a significant advancement, providing a data-driven supplement and extension to traditional risk analysis methods. This paper introduces a novel two-step rough fuzzy clustering method (TRF-BPC) for bankruptcy prediction, leveraging machine learning algorithms to enhance traditional risk management practices and emphasizing the need for sophisticated evaluation strategies to address risks. Firstly, the TRF-BPC algorithm innovatively employs rough fuzzy sets to process financial data, constructs boundaries (optimistic and pessimistic predictions) based on the correlations between attributes, and performs clustering by comparing the similarity between objects and these boundaries. Secondly, to more effectively handle uncertainties, a divide-and-conquer strategy is adopted for two-stage refined clustering. Additionally, the algorithm optimizes partition thresholds through grid search, avoiding the limitations of fixed thresholds. The method first processes and groups the attribute set, then calculates prediction values for objects based on these attribute groups. Subsequently, objects are classified by comparing their similarity with the prediction values. The effectiveness of this algorithm has been validated on nine datasets, with additional analysis conducted on its noise resistance capability. This attribute-oriented clustering process enables the analysis of factors influencing corporate bankruptcy, explores the interrelationships between these factors, and enhances the authenticity and validity of the clustering process. By dynamically assessing risks through stepwise clustering, this method addresses shortcomings in traditional financial risk analysis, such as data dependency, normal distribution assumptions, and static analysis. The TRF-BPC algorithm, with its robustness, accuracy, and adaptability, has emerged as a powerful tool for bankruptcy prediction and risk management in the evolving financial landscape.
This study explores the application of machine learning (ML) algorithms to predict lapses in investment policies, addressing a big challenge for insurance and financial services companies. The study compares three ensemble techniques: random forest (RF), gradient boosting (GB), and extreme gradient boosting (XGBoost), to identify the most effective model for predicting policy lapses and to determine the key factors influencing these predictions. The dataset used for this analysis is sourced from an anonymous insurance and financial services company on Kaggle, and includes data from 51,685 policies spanning from 2017 to 2020. Thorough data pre-processing, including handling missing values, outlier treatment, and feature scaling, is performed before training and evaluating the models. The results reveal that features such as tenure, number of missed payments, and total sum assured play a big role in predicting lapses. Random Forest is identified as the top-performing model. Furthermore, local interpretable model-agnostic explanations (LIME) is used to improve interpretability, offering detailed insights into feature contributions. These findings suggest that ML models, particularly Random Forest, are highly effective in predicting lapses in investment policies, offering valuable insights for insurance and financial services companies to manage and reduce policy lapses.
The adoption of blockchain technology (BCT) in supply chains presents a complex interplay of costs and benefits. While higher adoption rates can lead to significant reductions in costs, they also impose higher ongoing costs related to infrastructure, energy consumption, and technical maintenance. Through the utilization of system dynamics (SD) and regression methods, this study explores the intricate relationship between adoption rates, associated costs, and failure rates of blockchain implementation. Afterwards, a mathematical model is designed to optimize the adoption rate in order to reduce costs. The functions obtained from regression are integrated into this model, leading to the determination of the adoption rate. The results indicate that increasing the adoption rate may not always be profitable and could potentially increase costs. In this study, an adoption rate of 0.567 is obtained, suggesting that at this adoption rate, costs reach their minimum level, and that at higher or lower rates, costs increase.
The environmental, social, and governance (ESG) report is globally recognized as a keystone in sustainable enterprise development. However, current literature has not concluded the development of topics and trends in ESG contexts in the twenty-first century. Therefore, we selected 1114 ESG reports from global firms in the technology industry to analyze the evolutionary trends of ESG topics by text mining. We discovered the homogenization effect toward low environmental, medium governance, and high social features in the evolution. We also designed a strategic framework to look closer into the dynamic changes of firms' within-industry representiveness and cross-sector distinctiveness, which demonstrates corporate social responsibility and sustainability. We found that companies are gradually converging toward the third quadrant, which indicates that firms contribute less to industrial outstanding and professional distinctiveness in ESG reporting. Firms choose to imitate ESG reports from each other to mitigate uncertainty and enhance behavioral legitimacy.
This paper examines vertical outsourcing dynamics using three game-theoretic models: buyer's independent production, buyer-led outsourcing, and supplier-led outsourcing. The analysis systematically evaluates how government policy support affects outsourcing decisions, pricing strategies, and profit distribution in supply chains. Key results indicate that: (1) policy interventions significantly alter competitive dynamics, with buyer support encouraging price reductions to expand market share while supplier support enhances production efficiency and pricing power; (2) allocation of control rights directly determines profit distribution advantages, allowing the dominant party to secure higher profits through favorable contractual terms; (3) different policy types generate asymmetric profit effects, as buyer-oriented policies directly increase its profits whereas supplier-focused measures indirectly boost overall profitability through enhanced market competitiveness. Using Stackelberg game frameworks supported by numerical simulations, this research clarifies the dynamic interrelationships among policy interventions, control structures, and supply chain coordination, offering theoretical insights for industrial policy design and corporate outsourcing optimization.
Question-answering (QA) systems are vital for organizations to efficiently address customers' issues. Nevertheless, studies on QA systems in government are still lacking. Government QA systems face significant optimization challenges, including long response times, issue accumulation, and ineffective prioritization. This study develops an AI-driven system that leverages advanced BERT-based models to automatically classify issues, predict key attributes, and distribute them to appropriate departments using a novel matching algorithm. Trained on 812,322 citizens' inquiries from Messaging Borad for Leaders in China, our proposed system significantly reduces response times, minimizes uneven issue distribution, and decreases manual work and costs. The BERT-based models improve issue classification and attribute prediction accuracy by 5%-10% compared to baselines, while effectively reducing departmental overload. This study contributes to both organizational efficiency and New Public Management (NPM) literature by demonstrating how the government QA system's efficiency and service quality can be improved with the assistance of AI.
