Trinidad and Tobago is an energy-rich country and hence its revenue is highly dependent on oil and gas prices. Since its foreign income is primarily based on oil and gas sales, this heavy dependency significantly affects the exchange rates of various currencies with some more dependent on oil and gas prices than others. This study explores forecasting the monthly buying and selling rates of eight major currencies. Given the volatility of global energy markets, accurate forecasting is essential for economic planning and stability. The study incorporates energy-based predictors such as crude oil production, methanol exports and natural gas output, to improve forecast accuracy. A Random Forest Regressor identified the most influential indicators for each currency, followed by XGBoost, LSTM, and GRU models. Four ensemble models were created from a combination of LSTM, GRU, and XGBoost models. Rolling origin forecasting was used to replicate real-time scenarios and performance was evaluated using MAPE, MAE, and NRMSE metrics. Ensemble models performed significantly better than single models. Results confirmed that integrating energy indicators enhances predictive accuracy, offering valuable insights to central banks and policymakers managing exchange rates. We believe that this analysis will hold for similar energy-rich states.

The objective of this study was to determine the relationship between the Integrated Electronic Records System (SIRE) and tax compliance among VAT and corporate income taxpayers in Peru. The methodology used comprises: a quantitative approach, non-experimental design, and correlational scope, as the variables "Integrated Electronic Records System (SIRE)" and "Compliance with tax obligations" were defined. The technique used was a survey, with a 24-item questionnaire applied to a sample of 180 accounting department employees. The results indicate that 57.8% of respondents answered that SIRE is related to tax compliance. It was concluded that SIRE is moderately related to taxpayers' compliance with their tax obligations.

We describe a mobile application designed to enhance safety when using public transportation. During the training phase the application learns typical travel patterns and times by analyzing historical route data stored locally on the user's device. During operation, it identifies spatio-temporal anomalies in real-time by comparing a user's current location against their historical profile. If a significant deviation is detected, an alert is sent via SMS to the user's emergency contacts. All data and processing occurs strictly on the device itself and nothing is shared with the Cloud. It does not require an Internet connection (except for when it has to be installed in which case public WiFi can be used) making it useful for those who cannot afford cellular data plans. This paper provides a Proof of Concept of this application and includes typical use cases to illustrate efficacy.

Predicting the popularity of songs has long been a challenge in the music industry due to rapidly changing listener preferences and evolving production practices. This study applies machine learning techniques to audio features and temporal metadata spanning six decades of music to predict whether a track will achieve commercial popularity. Four models were developed and evaluated, including a tuned Random Forest with decade encoding, which demonstrated superior performance in terms of F1-score and ROC-AUC. Stratified sampling and per-decade validation ensured robustness against temporal bias, while feature importance analyses revealed danceability, instrumentalness, and acousticness as key predictors. SHAP-based explainability methods provided transparency into model decisions, enhancing trust and interpretability. A Streamlit-based application was also deployed to offer an interactive interface for exploring predictions, enabling threshold tuning and scenario testing. Results suggest that integrating temporal context with acoustic features improves predictive accuracy and provides actionable insights for playlist curation, artist scouting, and trend forecasting

Recent advances in deep learning offer new possibilities for evaluating academic writing, yet most systems assess essays as static texts rather than capturing development over time. This study proposes a hybrid Bidirectional Long Short-Term Memory (BiLSTM) and Convolutional Neural Network (CNN) model to assess writing progression in blended learning forums. The model is trained on benchmark datasets (TOEFL11, ASAP-AES). The system tracks writing improvement by integrating neural embeddings with interpretable linguistic features and provides insight into developmental trends. Results show strong performance on benchmark tasks (accuracy = 82.9%, QWK = 0.750) and successful generalization to forum data. A 17.6% average increase in predicted scores over 12 weeks demonstrates the model's sensitivity to real-world writing gains. This work contributes a scalable, interpretable framework for continuous writing assessment in online and blended learning environments.

Presently, emerging technologies like blockchain offer the potential to transform supply chains. The present paper explores unveiling supply chain performance (SCP) through blockchain adoption (BCA) and visibility mechanisms (transparency, trust, and traceability) in Pakistan. The study is quantitative, which proposes cross-sectional data. The context of the study is the logistics and transportation sector of Pakistan. The researchers would apply an online survey technique to get the responses from supply chain managers and logistics/operations managers. Using SEM analysis through AMOS, the study examine the impact of BCA on SCP and visibility (transparency, trust and traceability). The visibility (transparency, trust and traceability) is expected to be the substantial predictors of SCP. The results of study will assists policymakers and planners in developing an effective digital traceability system, real-time data sharing platforms, and secure transaction ledgers to make the SCP more effective and healthier. The study provides the guidelines to generate context-specific insights and digital transformation strategies for sustainable supply chain performance in emerging markets. The study overcomes the gaps in existing literature by offering an integrated framework which integrates BCA, visibility and SCP in a developing context.

Small and medium-sized enterprises (SMEs) are central to economic vitality but are stifled by the rigidities of centralized financial systems and operational inefficacies. Traditional models falter under operational constraints and informational imbalances, which are pervasive in supply chain finance (SCF). This study addresses these critical challenges by exploring the role of blockchain technology as a linchpin for innovation and efficiency in the SME sector. Utilizing a systematic literature review paired with grounded theory analysis, we identify the principal hurdles in SCF: operational bottlenecks, transparency issues, and regulatory constraints. The results emphasize that blockchain's capacity for bridging the gaps in conventional financing models can catapult SMEs toward sustainable growth and more robust market strategies. By facilitating a transparent, efficient, and secure SCF environment, blockchain stands out as a transformative force, signaling a paradigm shift in SME financial operations. The study concludes that blockchain integration is instrumental for SMEs to navigate the complexities of the modern economic landscape, indicating a significant leap towards a resilient and dynamic SCF ecosystem.

Traditional digital signatures depend on centralized authorities to build confidence, while blockchain mitigates this issue by distributing trust across a network. Consequently, the integration of digital signatures within blockchain systems becomes very interesting. The purpose of this review is to examine blockchain and digital signature convergence within academic studies. This work is based on the PRISMA framework to assemble studies stored in academic database, linking blockchain to digital signatures. Through this study, the number of publications associating blockchain with digital signatures has increased, reflecting a greater focus on the research scope. The study emphasizes the contributing countries and organizations, the most prolific authors, the relevant journals and publications, and the main research keywords and subject areas, recognizing their importance and the multidisciplinary nature of this research field. This paper proposes an additional research agenda to enhance the security and authentication of data, to develop strategies for the implementation of blockchain and digital signatures, and to enlarge the body of knowledge.

This paper proposes an innovative paradigm for evaluating stock behavior and performance using multi-step and multi-source sentiment signals within a deep RL pipeline that captures persistent investor sentiment dynamics that static, news-only approaches overlook. In contrast to conventional stock price analysis methods which are reliant on manual fundamental and technical analysis, our methodology harnesses the power of deep reinforcement learning algorithms - specifically DQN, PPO and A2C. We put forth a sentiment extraction framework that incorporates crucial financial documents like quarterly reports and newsletters with day-to-day news articles from various legitimate sources to reflect persistent fluctuation of investor and market participant sentiment. We designed a feature engineering process to create robust features for the Reinforcement Learning agent. Using historical SENSEX NSE stock data from 2021 to 2025, we demonstrate that our reinforcement learning framework excels in both return maximization and risk management.

The rise of generative artificial intelligence (Gen AI) is profoundly reshaping the innovation landscape, introducing new capabilities for creativity, collaboration, and strategic decision-making. As organizations increasingly adopt Gen AI tools to augment innovation efforts, there remains a significant gap in understanding how these technologies can be systematically integrated within open innovation ecosystems to foster sustainable, ethical, and inclusive innovation. This study addresses this gap by developing a strategy roadmap grounded in Sociotechnical Systems Theory (STS), offering a structured perspective on how Gen AI enables and enhances open innovation processes.

Sociotechnical Systems Theory provides a suitable lens for this investigation by emphasizing the co-evolution of technological infrastructures and human-centered practices. Within the context of open innovation-where knowledge flows across organizational boundaries-the interaction between technical tools and social structures becomes particularly critical. Gen AI, with its capabilities in automating ideation, enhancing real-time collaboration, and synthesizing complex data, operates at the intersection of these two subsystems. This paper investigates how Gen AI contributes to such interactions and what strategic pathways organizations can adopt to operationalize these contributions effectively.

To explore the enabling functions of Gen AI in open innovation, the study adopts an exploratory qualitative research design. Data were collected through 21 in-depth expert interviews and structured panel discussions using the Nominal Group Technique (NGT). Participants were selected from a diverse pool of European companies spanning various sectors-technology, manufacturing, healthcare, and consumer goods-and included both SMEs and multinational firms. This diversity allowed the research to capture a wide spectrum of perspectives, contextual variables, and implementation experiences. Additionally, experts represented different regions, enabling a richer understanding of how technological infrastructure, innovation policy, and cultural context influence Gen AI adoption and effectiveness.

The data from interviews and expert panels were analyzed using Interpretive Structural Modeling (ISM), a methodological tool well-suited for exploring hierarchical and interdependent relationships among qualitative variables. ISM enabled the researchers to identify twelve key enabling functions of Gen AI in open innovation and organize them into a multi-level framework that illustrates their interconnections and systemic influence. These functions range from foundational capabilities like workflow automation and productivity (WAP), to core operational activities such as idea generation and creativity (IGC), prototyping and co-creation (PCC), and knowledge management and sharing (KMS), to more strategic enablers like data-driven insights (DDI) and sustaining open ecosystems (SOE).

The resulting roadmap organizes these functions into five hierarchical levels, offering a clear representation of how foundational technologies support higher-order innovation outcomes. At the base, WAP provides the infrastructure necessary for digital efficiency. This function supports level-two activities such as stakeholder engagement (SEG) and cross-domain integration (CDI), which are critical for enabling inclusive collaboration and the synthesis of interdisciplinary knowledge-core pillars of open innovation. The third level encompasses innovation-centric activities, including IGC, PCC, real-time collaboration (RTC), market insights and commercialization (MIC), and ethical innovation management (EIM). These functions collectively form the operational engine of innovation ecosystems. DDI, at level four, links operational outputs to strategic insights, while SOE, at the apex, represents the long-term goal of building resilient, inclusive, and sustainable innovation ecosystems.

A key contribution of this study is the emphasis on the non-linear and systemic nature of innovation enabled by Gen AI. Traditional models of innovation often assume linear progressions from idea to execution. However, the roadmap presented in this study illustrates how feedback loops, indirect relationships, and cascading effects between functions are central to how Gen AI facilitates innovation. For instance, while EIM operates at an operational level, it has strategic implications for ecosystem sustainability and trust, aligning with recent concerns around AI ethics and transparency.

The study also highlights Gen AI's role in democratizing innovation by reducing participation barriers and enabling broader stakeholder involvement. However, it does not overlook challenges. The research recognizes that ethical, cultural, and technical complexities-such as algorithmic bias, data privacy, and organizational resistance-must be addressed to ensure the equitable implementation of Gen AI in open innovation frameworks. In this regard, the roadmap serves not only as a strategic guide but also as a diagnostic tool for identifying where interventions may be necessary to support responsible innovation.

From a theoretical perspective, the paper extends the literature on open innovation by integrating STS and offering a multi-level, interdependent model that reflects the realities of modern innovation ecosystems. It also contributes to emerging discussions around AI in innovation management by illustrating the specific functional roles of Gen AI and their interrelationships. The inclusion of EIM within the operational core adds a unique ethical dimension to the framework, building on recent calls to embed ethical considerations into AI-enabled systems from the outset.

Practically, the roadmap offers actionable guidance for managers and policymakers. Organizations can use the framework to assess their current readiness, identify priority areas for Gen AI investment, and develop phased implementation strategies that align with their innovation goals. Policymakers may also leverage these insights to support digital infrastructure, ethical governance, and cross-sector partnerships essential for scaling Gen AI-driven open innovation.

The purpose of this paper is to study the interactions between Strategic Foresight (SF) and Innovation Ambidexterity (IA) with the proposition of three mediating variables that could affect this relationship, namely: Knowledge Management (KM), Dynamic Capabilities (DC) and Perceived Organizational Support (POS). The current literature points to the existence of connections among SF, IA, KM, DC and POS. However, there are hardly any studies that analyze the dynamism inherent in these interactions and specifically how the combination of KM, DC and POS can explain the relationship between SF and IA from a holistic perspective. Based on selective review of prior literature on the main streams, and supported by theoretical bases rooted in resource based view RBV, SF, POS and DC theories, this research fills such a gap allowing, through an abductive reasoning, the development of 4 hypotheses forming our proposed framework. This study aims to contribute to advance the theories of foresight and innovation bodies by providing new insights on how the potential mediating variables proposed herein further enhance the links between SF and IA.

Artificial Intelligence (AI) is disrupting traditional markets in Latin America and the Caribbean (LAC) by enabling hyper-personalised, context-aware digital experiences that reshape customer engagement. This paper explores how AI-driven personalisation, when integrated with the Blue Ocean Strategy (BOS), creates entirely new business models tailored to underserved populations. By moving beyond saturated "red oceans," LAC firms are tapping into uncontested "blue oceans" of market growth through value innovation. The study highlights AI technologies such as machine learning, natural language processing, recommendation engines, predictive analytics, digital twins, and agentic AI, showcasing how they enable scalable and emotionally intelligent user experiences. Through real-world case studies in e-commerce, education, and healthcare, including Magazine Luiza, UnaAI, and 1DOC3, we demonstrate how AI personalisation drives inclusion, accessibility, and customer retention by adapting to regional cultures and digital readiness levels. Emerging technologies like digital twins and agentic AI further shift personalisation from reactive to proactive, allowing simulations of user behaviour and autonomous decision-making aligned with cognitive science principles. These innovations not only eliminate frictions and reduce operational costs but also raise relevance and create novel, trust-driven interactions, core tenets of BOS's ERRC grid. While ethical challenges remain particularly around data privacy, bias, and infrastructural disparities, the paper presents design and policy recommendations to ensure culturally adaptive and equitable AI deployment. Ultimately, by leveraging AI as a strategic tool for market creation, LAC firms can pioneer disruptive business models that deliver human-centered, sustainable innovation across sectors.

This study proposes a patent-based framework to measure the structural integration of artificial intelligence (AI) as a General Purpose Technology (GPT). Using global patent data from 1980 to 2020, we classify AI-related patents into Core AI and applied AI categories based on CPC codes and keyword extraction. The analysis reveals increasing cross-sectoral diffusion of AI, with notable variation across countries. This approach captures the depth and breadth of AI's technological embedding, offering a complementary tool to existing AI indices for assessing national AI capabilities.

Developing a Shared Vision is a critical yet underexplored challenge in the management of complex, inter-organizational Research, Development, and Innovation (RDI) projects. While early-phase alignment tools exist, few methods support shared sensemaking at the project's midpoint -- when shifting goals and emerging insights demand renewed alignment. This paper presents and evaluates an Experiential Analogy to Mid-Project Visioning, tested within a European rail innovation project. A Cooking-based Workshop was designed to help diverse consortium members reflect on project dynamics through a familiar, hands-on experience. The approach integrated Analogical Reasoning, Design Thinking tools, and Experiential Learning to spark creative insight, foster alignment, and enhance collaboration. The intervention produced three main outcomes: (i) a visual representation of the Shared project Vision; (ii) Strategic Insights that reinterpreted key project components; and (iii) Design Knowledge to inform subsequent development. The results underscore the potential of analogy-based experiences to support mid-course alignment, particularly in RDI settings. Future research should examine the applicability of alternative analogies to validate whether metaphors beyond cooking can yield similar benefits. It should also explore the benefits of integrating creative, experiential and empathic methods - such as those rooted in Design Thinking - into Project Management (PM) practices. This study contributes to PM literature by offering a novel, Human-Centered approach to fostering mid-project visioning in inter-organizational innovation settings.

This paper aims to study the impact of Strategic Foresight (SF) on Innovation Ambidexterity (IA) and investigates the effects of Knowledge Management (KM), Dynamic Capabilities (DC) and Perceived Organizational Support (POS) in influencing this relationship. While existing literature highlights connections among SF, IA, KM, DC and POS, there are - to the best of our knowledge - no empirical studies that analyze the dynamic interplay among these constructs and particularly how the joint effects of KM, DC and POS can interact together to leverage the SF-IA relationship. To empirically test the hypothesized effects, we draw on a survey data collected from 76 foresight practitioners. The findings show that SF has no direct impact on Innovation ambidexterity, while KM, DC and POS are mediating variables that jointly account for this relationship. Theoretical contributions and managerial implications along with future research directions are also discussed in this paper.

The objective of this study was to determine the relationship between the use of the Peruvian tax administration's digital platforms by doctors to file and pay their taxes and the collection of taxes from the income of these professionals, as well as the difficulties they face in complying with their tax obligations. This study has a quantitative approach, a non-experimental design, and a descriptive explanatory scope. A structured questionnaire was administered to a sample of 380 doctors selected based on the inclusion criteria of being from Lima and licensed to practice medicine. The results show that 33% have limitations in using the tax administration's digital platforms to file and pay their taxes, and 68% have little knowledge of these platforms for filing their tax returns. It is concluded that there is a negative relationship between the use of the digital platform and the collection of taxes from the income received by doctors, influenced by a lack of knowledge about the use of digital tax platforms, a lack of attitude and/or motivation to pay taxes, and a poor understanding of tax legislation.

India's agriculture sector is vital for employment and the economy, but it faces several challenges despite digital initiatives. The agricultural sector is unorganized and fragmented, lacking integrated social networking and hindering collaboration. This study employs a qualitative, exploratory design using semi-structured interviews with ten agri stakeholders. The thematic analysis identified four key barriers: lack of reliable and accessible information channels for agricultural knowledge, weakening institutional support and trust, unstable income generation, agricultural market inefficiencies, and farmers' disempowerment. This study contributes a framework for a dedicated digital platform enabling peer-to-peer learning and empowering farmers through unified stakeholder engagement.

In the pixel-oriented world where one's imagination turns to reality, giving a 360-degree view of the environment, Virtual Reality (VR) is an emerging technology for Tourism in India, but when glanced towards the global world, many takeoffs need to be discovered. This study aims to unveil potential opportunities in VR tourism and compare the practices being followed by Dubai (UAE) and India to accelerate the tourism outlook in the context of Virtual Reality. This study examined several government websites operated by the region, facilitating 360° overviews. A comparison was made based on several evaluation frameworks.

The bibliometric study we have carried out analyses the evolution of scien-tific research on intelligent educational technologies in the specific context of vocational retraining. The study covers a period of 20 years (2004-2024). After filtering, the work was carried out based on 164 publications, all in-dexed in Web of Science. Through this analysis, we identify the main trends around our research axes, the most influential groups of authors, the topi-cality of the subject, as well as emerging themes in this rapidly expanding field. Our study has considered the COVID period, with its e-learning con-straint, which has had a major impact on publications on the subject we have addressed. These results underline the growing importance of this field of research. Our findings include a significant increase in the number of publications, with an annual growth rate of 13.23%, a lack of longitudinal studies, and the fact that most studies fail to take account of the profession-al and cultural reconversion contexts of underdeveloped countries.

Engineering Mechanics Dynamics is an essential subject in engineering education. However, many students struggle with this subject, resulting in a significant number of academic difficulties. These challenges often arise from difficulties in visualizing static images, comprehending complex engineering models, and misunderstanding key concepts. The objective of this research was to understand how a Metaverse (virtual) space can be used to support the communication, collaboration and problem solving in an academic engineering lab. Two separate groups were assessed using a controlled experimental method. One group took part in traditional, face-to-face group discussions, while the other engaged in discussions within a real-time virtual environment. The study produced significant findings, highlighting a difference in academic achievement between the two interaction methods. Additionally, participant feedback, gathered via the User Experience Questionnaire (UEQ), revealed positive user experiences, scoring favorably across all six UEQ scales. Users found the system intuitive, efficient, and effective. The real-time virtual environment demonstrated higher academic success, enhancing both the effectiveness and user experience (UX) of the Metaverse platform in engineering.

This systematic review offers an in-depth and crit- ical examination of transfer learning and domain adaptation strategies as applied to the semantic segmentation of 3D cultural heritage data. The inherent complexity, variability, and hetero- geneity of 3D heritage datasets often hinder the effectiveness of conventional segmentation models in terms of precision and generalization. To address these limitations, transfer learning and domain adaptation have gained prominence by enabling the reuse of knowledge from pre-trained models and mitigating domain shifts between source and target data. This review systematically categorizes and evaluates existing methodologies, with a particular emphasis on their application within the domain of cultural heritage. It encompasses a broad spectrum of topics, including 3D data acquisition techniques, the design and refine- ment of deep learning architectures, and the metrics employed for performance evaluation. Furthermore, the review discusses persistent challenges such as limited labeled data, variability in lighting conditions, material textures, and geometric scales. By integrating recent advances and synthesizing key contributions across the literature, this work aims to provide a foundational reference for researchers and practitioners seeking to improve the semantic segmentation of 3D cultural heritage objects and environments.

The pervasive adoption of AI in modern workplaces has created a paradox: while promising efficiency, it often imposes unsustainable cognitive challenges on employees. To create a balance system between AI capabilities and human experience, this study investigates how organizations can harness AI's potential without sacrificing employee well-being. This study pioneers a dual-theoretical lens integrating the Stimulus-Organism-Response (SOR) framework with Cognitive Load Theory (CLT) Through a qualitative study relying on in depth interviews with 10 professionals across knowledge-intensive sectors in Morocco. The findings reveal that AI adoption presents some challenges characterized as technical and emotional barriers that act as an external stimulus triggering employee's cognitive overload serving as an internal organism. This in turn alarms organizations to set some best practices acting as a collective, adaptive response to manage and reshape the emerging cognitive burdens. This study contributes to the existing literature by repositioning the SOR framework as a proactively manageable system reframing the response component as a structured, managerial effort to manage AI-induced cognitive demands and build a sustainable human-AI co-functionality. For practitioners, this study equips organizations with a three-pronged roadmap for a human-centric AI integration.

Paradigm-shifting technologies (PT), such as quantum computing (QC), rely on novel scientific principles and present substantial potential for future applications. Because their development is both complex and costly, identifying long-term use cases that serve as guiding principles for technology development is crucial. However, focusing solely on corporate-level scenarios may yield insufficient returns, overlooking broader socioeconomic benefits that could extend across entire industries or even national economies. Yet, a systematic method to align the innovative capabilities of PT with socioeconomic goals remains absent. This paper introduces a conceptual framework guiding potential technology adopters to map a socioeconomic domain for PT application identification. First, broad innovation fields are defined based on public innovation systems. Second, these innovation fields are enriched with socioeconomic needs using the United Nation (UN) Sustainable Development Goals (SDGs). Ultimately, this framework derives specific socioeconomic development needs that underscore the wide-ranging impact and value of PT in advancing strategic goals.

The tradeoff between internet of things (IoT) network growth and user requirements significantly maximized which affected by the dynamically devolving malware attacks. The conventional security systems have utilized for scalable and intelligent data security, especially intrusion detection systems have used for IoT network by using the different network parameters. Several recent solutions also utilized for IoT network to address the problems, this work present the cryptography with transfer learning model for malware detection in IoT network. Here, the dynamic chaos encryption model is used for encryption purpose which generates the encrypted data with proper key generation process to ensure the data protection. The transfer learning model evolved from the fine-tuned rule-set includes convolutional neural network (CNN) and recurrent neural network (RNN) is used for network traffic patter analysis and detect the malicious types in the IoT network. The proposed work utilizes the Bot-IoT and TON_IoT datasets to validate the performance. The Chaos+RNN model achieved 99.624% accuracy on the Bot-IoT dataset, outperforming the previous best by 1.272%, with precision, recall, and specificity improvements of 1.301%, 1.308%, and 1.055%, respectively. On the TON_IoT dataset, it reached 99.472% accuracy, exceeding the top model by 1.365%, with respective gains of 1.447%, 1.449%, and 1.168%. The findings show that the system achieves superior malware detection performance, with increased precision and recall across various attack scenarios.

This study explores the roles and challenges of virtual influencers (VIs) in the insurance sector, a field characterized by trust, regulation, and informational complexity. Grounded in Source Credibility Theory and the Persuasion Knowledge Model, the study draws on a qualitative analysis of professional and consumer perspectives to examine how VIs are perceived and potentially integrated. The findings indicate that receptiveness to VIs is shaped by organizational readiness, communication culture, and ethical considerations. While VIs offer potential for simplifying information and modernizing brand image, their use remains limited to supportive, non-advisory roles.

This systematic review analyzes the impact of digital marketing strategies on the increase of tourism between 2018 and 2024. The aim is to assess how digital strategies influence tourism promotion through theoretical and empirical studies. A total of 30 articles in English and Spanish were reviewed, sourced from databases like Scopus, Web of Science, and ProQuest, using keywords related to digital marketing and tourism. The results indicate that digital marketing has boosted tourism, with strategies such as social media, Earch Engine Optimization and content marketing standing out. Studies like those by Herrera et al. (2019) and Sarquis et al. (2015) show that these strategies increase visitor conversion and improve the perception of tourist destinations. The integration of advanced technologies and personalization are key factors for success. In conclusion, digital marketing is essential for increasing tourism when implemented strategically.

The objective of this study was to determine the relationship between digital marketing strategies and customer loyalty in an entertainment company in Trujillo in 2024. A questionnaire was administered to 371 customers of the entertainment company. The results showed a Spearman correlation coefficient of 0.824 and a bilateral significance of 0.000 (ρ < 0.05), indicating a strong direct correlation between digital marketing strategies and customer loyalty. Additionally, it was demonstrated that flow, functionality, and feedback are significantly related to customer loyalty, with coefficients of 0.692, 0.727, and 0.720 respectively, and a bilateral significance of 0.000 (ρ < 0.05). The study concluded that the correct and timely implementation of digital marketing strategies significantly enhances customer loyalty levels.

In recent years, the integration of sustainability principles into digital marketing strategies has become increasingly important, particularly within the textile and fashion sectors. However, conventional machine learning approaches often treat input features as black-box predictors, neglecting their causal relevance. In this study, a novel framework called CATE-Informed Engagement Prediction Framework (CIEPF) is proposed. This framework enhances predictive models by incorporating estimated CATE (Conditional Average Treatment Effects) as auxiliary features. By embedding individualized causal estimates into the feature space, models are enabled to capture nuanced treatment-response heterogeneity that standard descriptive variables typically overlook. To validate the framework, a dataset of Instagram posts related to textile circularity has been analyzed across three treatment conditions: positive sentiment, afternoon posting, and weekend posting. A comparative evaluation was conducted using Random Forest and Decision Tree models, both with and without CATE integration. The empirical findings demonstrate that models enriched with CATE consistently outperformed their baseline counterparts, achieving lower RMSE and higher R^2 values. These results confirm that the inclusion of causally informed features significantly enhances predictive accuracy in engagement modeling. The proposed methodology bridges the gap between causal inference and supervised learning, offering valuable implications for personalized digital marketing and sustainable communication strategies.

This paper introduces a new consumer identity concept, "I Consume Myself," which captures how individuals increasingly become both the subject and object of consumption in digital and crisis-driven environments. Drawing on a three-year longitudinal study that combined surveys, interviews, and phenomenological analysis, the research demonstrates how global crises and virtual platforms accelerate the transformation of consumer behavior. Findings reveal that consumers construct "virtual selves" shaped by algorithms and social media, while simultaneously consuming their own identities through self-branding, image management, and digital performance. The paper situates this concept within the broader framework of the virtual consumer, extending existing theories of consumer identity and postmodern consumption. The study contributes by defining and empirically grounding the "I Consume Myself" identity, highlighting its implications for innovation, marketing, and strategic management in the digital era.

Artificial Intelligence (AI) has proven to be a force of revolution in various industries, reshaping conventional paradigms in terms of efficiency, precision, and novel capabilities. Health is one such industry where the influence of AI is both extensive and multifarious. Be it automating mundane tasks or aiding intricate clinical decisions, AI technologies are revolutionizing the mode, management, and optimization of care. This bibliometric analysis analyzes the new literature on the application of AI to hospital and healthcare systems. It aims to uncover the evolution of research patterns, identify leading authors, institutions, and countries working in this field, and map scientific collaboration patterns through co-authorship networks. Ultimately, the findings not only highlight the core role of AI to enhance hospital performance and patient care but also point toward potential pitfalls and research opportunities for building intelligent, adaptive, and fair healthcare systems

Early detection of Parkinson's Disease (PD) is critical for timely intervention and effective disease management. Dopaminergic imaging techniques, including Single Photon Emission Computed Tomography (SPECT) with ¹²³I-Ioflupane, have shown significant promise in detecting Parkinson's Disease during its early stages. Nonetheless, the existing dependence on manual interpretation leads to variations among observers and inconsistencies in diagnosis. This research introduces an automated model aimed at the early detection of Parkinson's Disease, utilizing image processing techniques alongside an Artificial Neural Network (ANN) to overcome existing limitations. A collection of 200 SPECT images consisting of 100 from healthy individuals and 100 from patients with Parkinson's disease was obtained from the Parkinson's Progression Markers Initiative (PPMI) database. Image processing techniques were utilized to delineate the Caudate and Putamen regions, which serve as critical areas of focus in the diagnosis of Parkinson's disease. The area values obtained were subsequently fed into a custom-designed artificial neural network aimed at emulating human pattern recognition in image-based diagnosis. The proposed ANN exhibited impressive performance, attaining an overall classification accuracy of 94%, alongside a sensitivity of 100% and a specificity of 88%. The findings indicate that the established system has the potential to function as a dependable and effective decision support resource for healthcare professionals, possibly improving the precision and uniformity of PD diagnosis in clinical settings.

Despite the rapid growth of mobile healthcare apps, many vulnerable populations such as older adults, those with cognitive impairments, and people in under-resourced communities are left behind due to limited digital literacy, outdated technology, or unreliable internet access. This paper introduces a mobile application designed to overcome these very challenges by simplifying medication management to foster independence. Our system reduces the cognitive and technical burden on users by allowing them to simply scan prescription labels, which automatically extracts crucial information, tracks inventory, and records dosage histories. It effectively manages stock by organizing medications based on expiration dates and streamlining the transition to new supplies. Built with Flutter and Dart for cross-platform support and FastAPI for backend operations, the app is optimized for low-resource environments and offline use. Ultimately, this solution aims to empower vulnerable populations, giving them greater confidence, autonomy, and efficiency in managing their own healthcare.

Existing colorectal cancer (CRC) detection practices have historically relied on manual examination and conventional imaging, hindered by subjectivity and variability. Recent advancements in artificial intelligence (AI), including convolutional neural networks (CNNs) and transformer-based models, have enhanced automated polyp detection, yet challenges such as variable polyp sizes, poor lighting, and noisy colonoscopy backgrounds persist. Early CRC detection is pivotal, significantly boosting survival rates and reducing the need for invasive treatments through prompt polyp identification and intervention. Accurate and efficient segmentation of colorectal polyps from colonoscopy images is essential for advancing CRC outcomes. We present an innovative DeepLabv3+ segmentation framework, integrating channel-spatial attention modules, attention gates, and cascaded dilated convolutions, complemented by a boundary-aware loss function. This architecture tackles the complexities of variable polyp sizes, suboptimal lighting, and background noise across public datasets: Kvasir-SEG, CVC-ClinicDB, and ETIS-Larib. Our model achieves Dice coefficients of 0.924, 0.912, and 0.881, respectively, surpassing or matching the outcomes of existing CNN and transformer- based approaches which are regarded as the standard in this use case. The incorporation of attention mechanisms sharpens feature extraction by focusing on polyp boundaries, while the boundary-aware loss refines edge precision. An ablation study confirms the efficacy of these components in enhancing segmentation performance. This robust medical image segmentation advancement offers potential for large-scale CRC screening and diagnosis. By enabling precise polyp delineation, it supports early detection, equipping clinicians with actionable insights for confident decision-making. The model's precise boundary accuracy and attention-driven analytics strengthen intelligent decision-support systems, optimizing treatment strategies and elevating patient care.

In the finance and banking industry, product recommendation systems play an important role in enhancing customer engagement and driving cross-sell and upsell opportunities. However, traditional recommendation models often fail to account for product eligibility constraints and customer-specific suitability, leading to irrelevant or infeasible suggestions. This paper introduces a novel hybrid framework that combines rule-based eligibility filtering, customer segmentation, and machine learning to generate personalized, actionable product recommendation. The goal of this paper is to predict the most suitable financial product for a customer by mimicking human reasoning. To achieve this, our research employs a hybrid approach that combines fuzzy logic rule-based eligibility filtering with a supervised machine learning model to estimate adoption probabilities and recommend the next best product for the customer. Key features such as income, risk, credit score, and employment status are used together with other engineered features such as customer segment, balance to income ratio and eligibility scores to rank products for each customer. A weighting mechanism is introduced to prioritize recommendations that align with both customer needs and business constraints. Evaluation of synthetic data generated to mimic real-world data shows that the incorporation of eligibility and behavioral characteristics significantly improves the relevance of the recommendation and predictive performance. This system presents a scalable and interpretable solution for financial institutions that aims to optimize customer targeting while maintaining operational feasibility. Ideally, recommending the right product to the right customer.

As enterprise attack surfaces grow in complexity and scale, traditional cybersecurity practices struggle to keep up with the velocity, sophistication, and automation of modern threats. This paper presents a deep technical analysis of how artificial intelligence (AI), machine learning (ML), large language models (LLMs), and agentic AI can transform critical functions in enterprise cybersecurity from threat detection and incident response to identity, data, and vulnerability management. The paper explores AI driven techniques to enable proactive defense, threat triage, policy generation, and autonomous agents capable of orchestrating detection, decisioning, and remediating in near real-time.

Pipeline leaks in water distribution networks cause major losses and infrastructure damage. While traditional detection methods are effective at locating leaks, they remain costly and unsuitable for real-time monitoring. This study develops a machine learning framework for leak detection and localization using pressure data enhanced by signal and statistical feature engineering. Signal transformations (Fourier, Wavelet, Cepstral), statistical metrics, and three classifiers (XGBoost, SVM, and ANN) were evaluated on a benchmark dataset under varied flow conditions. Post optimization, the ANN achieved 97% accuracy under low-flow scenarios, with XGBoost reaching 93% and demonstrating superior robustness. SVM matched peak performance but lacked consistency. A meta-analysis showed this study's ANN, using only pressure data, outperformed a comparable high-flow condition pressure-only model (77% vs. 71.8% accuracy). However, the comparative model using multiple sensor types achieved higher accuracy (86.5%). Nevertheless, the XGBoost model of this research surpassed all benchmarks (89%) due to the richer set of features. Feature importance analysis highlighted the discriminative role of dominant frequencies, wavelet coefficients, and statistical descriptors. Overall, the findings confirm that integrating signal analysis with statistical profiling enables scalable and accurate leak detection in smart water networks.

The adoption of Artificial Intelligence (AI) in the air cargo industry holds immense potential to enhance operational efficiency, decision-making, and competitiveness. However, its implementation is significantly hindered by various technological barriers. This study employs the Fuzzy DEMATEL methodology to identify, evaluate, and analyse the interrelationships among these barriers. A total of 14 key technological obstacles were examined, and the analysis revealed that factors such as poor R&D on AI adoption, lack of standardisation, and erroneous AI algorithms act as primary causal barriers, exerting strong influence over the system. In contrast, barriers such as low AI maturity, uncertain progression, and integration issues emerge as effect barriers, shaped by upstream technological deficiencies. The findings provide a strategic framework for stakeholders to target high-impact drivers to effectively mitigate systemic resistance. The study offers practical insights for technology managers and policymakers aiming to foster AI readiness and resilience within the evolving air cargo ecosystem.

Generative AI (GenAI) is transforming industries and business in rapid fashion, offering unprecedented opportunities for automation. Organizations are often burdened by administrative tasks, such a record keeping. Record keeping is essential for accountability, care quality, and regulatory compliance, yet professionals often experience it as time-consuming and burdensome, diverting attention from client engagement. To address this challenge, the aim is to test in practice how an GenAI powered app can enhance the record keeping capability of an organization.

This research was conducted at the Dutch Salvation Army's Juvenile Protection department in the Netherlands. Following an action research approach, we tested the Luisterlinie application, a GenAI-powered tool designed for transcription and summarization of client conversations and meetings. Over seven weeks (April-June 2025), 47 youth protection professionals participated in the pilot. Data were collected through interviews, three surveys, application usage logs, and feedback sessions. Quantitative and qualitative analyses were combined to assess impacts on time spent, quality and completeness, and, and user satisfaction.

Findings demonstrate that GenAI reduces record keeping time, ease administrative workload, and improve user satisfaction, while enabling professionals to remain more focused on client interaction. Participants reported perceived efficiency gains and similar completeness as compared to humans. However, challenges emerged regarding transcription accuracy, speaker attribution, and integration with existing workflows, underscoring the continued need for human oversight.

The study concludes that GenAI applications like Luisterlinie can enhance organizational record keeping capability, particularly in contexts where minor inaccuracies are tolerable. While promising, these tools should complement rather than replace professional judgment. Broader adoption requires improvements in accuracy, workflow integration, and reliability.

One of the most important and promising areas for the application of generative AI and large language models (LLMs) is customer relationship management (CRM). These emerging technologies greatly enhance the automation of complex CRM processes by improving customer interactions and personalization. This study, based on articles from the Scopus database published between 2000 and 2024, examines research on generative AI and its impact on CRM, providing a comprehensive overview of global advances in the field. In addition, the study identifies underexplored areas and highlights a significant gap between existing research and potential applications of generative AI in CRM. This underscores the need for further empirical investigations, supported by robust theoretical frameworks, to enrich our understanding of the impact of generative AI on CRM. The study also opens up new avenues for future research, particularly in the integration of generative AI at different stages of CRM development. It advocates the creation of methodologies and tools to improve customer relationships, thus bridging the gap between theoretical knowledge and practical applications.

The objective of this study was to evaluate the purchase intention of functional products made from tarwi among parents in the district of San Martín de Porres, Lima, in 2025. The study was quantitative, basic, descriptive, and non-experimental in design. A structured survey based on a five-point Likert scale was administered to a sample of 175 parents from a private school in the district of San Martín de Porres. The dimensions analyzed were: purchase interest, social influence, and perceived control. The results show that the majority of respondents have a favorable attitude toward the consumption of products made from tarwi, especially when these are perceived as beneficial to their children's health, accessible in the local environment, and endorsed by trusted figures such as health professionals. It concludes that there is significant potential to position tarwi as a functional food in urban contexts such as San Martín de Porres. These findings contribute to the understanding of consumer behavior around traditional foods with added value and can serve as a basis for future sustainable business development and food education initiatives.

This systematic review synthesizes empirical evidence on how Artificial Intelligence (AI) shapes corporate customer experience (CX) across industries between 2021 and 2025. Following PRISMA procedures, we screened peer-reviewed studies indexed primarily in Scopus and complemented with reference chasing. Forty-one articles met the inclusion criteria. The synthesis indicates that AI augments CX through hyper-personalization, conversational interfaces chatbots and voice assistants, recommendation and prediction engines, and real-time analytics that orchestrate omnichannel journeys [1], [10], [13], [32], [35]. Benefits cluster around higher satisfaction, trust, convenience, conversion, loyalty, and lower churn; yet outcomes depend on transparency, explainability, data quality, ethical safeguards, and cultural fit [7], [10], [11], [32], [38]. Boundary conditions include sector banking, retail, hospitality, public services, task complexity, customer digital literacy, and perceived risk [6], [12], [31], [39], [41]. We propose an integrative framework in which AI capability with perceived usefulness, personalization, effort reduction and CX outcomes, moderated by trust, privacy control, and service context. Managerial and policy implications and an agenda for future research are discussed.

This study employed a methodological design oriented toward achieving a clear understanding of the variables that condition the influence of the Marketing Mix on sales performance at La Casa del Furgón, a commercial enterprise based in Jaén. The research was conducted under a non-experimental scheme, in which participants were examined within their natural context, avoiding deliberate intervention in the process. A cross-sectional and correlational approach was adopted, which enabled the analysis of data at a specific moment while identifying the degree of association between the variables under study. The population consisted of 110 customers, selected through a random sampling technique, who provided relevant information via structured surveys and questionnaires. The results revealed a significant and positive correlation between the dimensions of the Marketing Mix and sales levels, validated by the corresponding statistical coefficients. Likewise, linear regression facilitated the development of a predictive model that explained the relationship between the variables, confirming the direct effect of decisions associated with each component of the Marketing Mix on sales performance. These findings are consistent with previous research that emphasizes the importance of continuously improving marketing strategies to strengthen competitiveness and stimulate sales growth. In this context, it is advisable that La Casa del Furgón periodically evaluate the traditional 4Ps (product, price, place, and promotion) and complement them with the 4Es (experience, exchange, evangelism, and everywhere) as an alternative perspective to respond effectively to the evolving demands of consumers. The study provides a robust empirical basis for strategic decision-making in the field of marketing management. The evidence obtained contributes practical insights that may support both the consolidation of business success and the pursuit of sustainable growth within increasingly dynamic and competitive markets.

Research often considers the use of information and communication technologies (ICTs) as a detrimental factor that negatively affects individuals' well-being and productivity. However, a growing number of studies have started to highlight the benefits of non-work-related ICTs' contribution to work recovery, through replenishing depleted energy and promoting additional vigor. ICTs could lead to improved health and work outcomes. To enhance the understanding of non-work-related ICTs' impact on the individual's work recovery, we undertake a systematic review of the research covering both at-work and off-work contexts. Specifically, we review 50 empirical studies and consider the recovery process and its antecedents and outcomes. Within the recovery process, we examine the impact of specific non-work-related ICT activities (e.g., social media, online videos, digital games, and non-clinical online interventions) on the four most researched recovery experiences, namely psychological detachment, relaxation, mastery, and control. The review concludes by putting forward three promising areas to guide future research. It offers both theoretical and practical implications by clarifying how non-work-related ICT activities influence work recovery, and by providing actionable guidance for individuals and organizations to enhance health and work-related outcomes.

This study systematically analyzes how entrepreneurship influences job creation worldwide between 2021 and 2025. We searched Dialnet, ProQuest, Scielo, and Web of Science, applying inclusion/exclusion criteria and the IMRaD structure. Forty studies met the criteria: 32 scientific articles and 8 systematic reviews. Findings show that entrepreneurship is a strategic lever for employment generation through startups, SMEs, and social ventures, with effects moderated by policy, finance, education, and digitalization. The employment impact is heterogeneous across contexts but robustly positive overall. Implications include strengthening entrepreneurial ecosystems to scale quality and sustainable jobs.

Traditional supply chain risk management frameworks consider disruption events in the supply chain as sources of cascade effects that impact the operating performance of the focal company. There is evidence that also Product compliance issues can generate supply disruptions and thus its consideration shall be integrated in comprehensive supply risk management approaches. This paper proposes a stepwise method for assessing Product Compliance Business interruption risk (PCBIR) due to material, supplier, and supply chain risk factors and integrates them into a comprehensive Supply Risk Assessment framework. The likelihood of noncompliance originated by the supplier is estimated by considering different event types and the related conditional probabilities. The related impact on the focal company's business is assessed including the expected recovery times for the different compliance events and the affected final sales. A pilot application was performed involving a multinational company in the construction industry to demonstrate the feasibility of the proposed method. The paper contributes to extending the supply risk management knowledge for tackling product compliance issues in complex global supply chains. By applying the proposed method, managers can identify and assess suppliers' PCR and make informed decisions for effective risk management strategies.

The transition of the automotive industry towards sustainability is critically influenced by the drivetrain technologies employed in various vehicles. This study aims to determine which technologies are the most environmentally favourable under various conditions, particularly in the context of the global shift towards reducing greenhouse gas emissions. A systematic literature review (SLR) was conducted, drawing insights from diverse studies to evaluate the environmental performance of different propulsion systems. The analysis reveals that Battery Electric Vehicles (BEVs) are predominantly the most eco-friendly option across most scenarios, especially when powered by electricity derived from renewable sources. Fuel Cell Electric Vehicles (FCEVs) also present a viable alternative when hydrogen is produced via carbon-free methods. In regions where reliance on fossil fuels persists, Internal Combustion Engine Vehicles (ICEVs) remain relevant, while Compressed Natural Gas (CNG), especially when sourced from biomethane, is recommended for trucks due to its ecological benefits. The overarching conclusion emphasizes that the environmental impact of these technologies is significantly shaped by the energy sources powering them. This research provides a foundational understanding necessary for driving policy and industry decisions towards low-carbon transportation solutions.

Artificial Intelligence (AI) is increasingly adopted in critical sectors such as healthcare, finance, and public administration, where it promises significant gains in efficiency, automation, and decision support. At the same time, these systems expose societies to serious risks, including bias, discrimination, safety failures, and privacy infringements. To document and learn from such failures, the Artificial Intelligence Incident Database (AIID) was created as a community-driven repository and collective memory of AI harms, designed to support research, best practices, and governance. As of 2025, the AIID contains more than 1,100 incidents, yet its unstructured, narrative reports make systematic analysis difficult and limit their policy value. This paper addresses that challenge by applying a Large Language Model (LLM) pipeline, guided by the OECD AI Incident Reporting Framework, to transform AIID reports into structured data and enable systematic analysis of recurring patterns in AI incidents. The analysis reveals a sharp rise in frequency and severity since 2020, with human and economic harms dominating, transparency and fairness most frequently violated, and ICT, finance, and public administration accounting for most cases. In terms of implications, the study provides insights to better inform AI governance. It also demonstrates how to use LLMs to transform unstructured dataset into structured data for analysis.

Intermittent demand data, characterized by frequent zero values and unusual patterns, pose a distinct challenge for conventional and even advanced forecasting approaches. This paper introduces a novel framework that repurposes a lightweight GPT-2 model for forecasting in inventory-related settings, where demand signals are of an intermittent nature. The approach eliminates tokenisation by mapping raw numerical values directly into the Large Language Model (LLM) latent space and augments the decoder with an optional zero-inflated head to handle the high frequency of zero sales data. Experiments on three real-world Stock Keeping Units and four forecast horizons benchmark the proposed model against seven classical and machine learning (ML) baselines. The LLM reduces the Root Mean Squared Scaled Error by up to 27% compared to the best competing method in both short (7-day) and long (42 / 49-day) windows while remaining competitive in intermediate horizons. Complementary error metrics corroborate these gains. A preliminary pattern-recognition study links higher LLM improvements with greater intermittency and stronger chaotic signatures (mean Lyapunov exponent). These results establish that a properly fine-tuned, decoder-only LLM can serve as an accurate and adaptable forecaster in forecasting scenarios previously dominated by both classical intermittent-demand models and modern ML approaches such as LSTMs and gradient-boosted trees.

Stakeholders, through their vested interest in a project hold the key to project success. Therefore, the management of them is crucial and mandatory. However, tasks involved in stakeholder management is not that simple and thus require tools to perform them. The challenge is that some of the existing methods, techniques and tools are incapable to fulfil the performance of these tasks. The evolvement and sophistication of project stakeholders do not make the situation easier too. In an attempt to address this challenge, researchers have developed software tools for effective and efficient of management stakeholders. Artificial Intelligence (AI) is a technology that permeates and revolutionizes almost every business sector in terms of automation of tasks and efficiency improvement. AI has also found usage in various areas in project management. Although there is a wealth of research on the benefits of AI, little is known about how using AI tools has assisted project managers to manage stakeholders better. This study sought to address this research gap by conducting a systematic mapping review. The study found that AI has benefited the stakeholder management knowledge area by improving stakeholder identification, analysis, classification and risk management process. However, the study also established that there is a need for much research in certain research categories such as, evaluation, proposal solution, philosophical, opinion and experience papers pertaining stakeholder management.

The growing capabilities of LLMs and AI agents offer new opportunities for enhancing Technology Roadmapping, a core method in strategic planning. While current applications of LLMs in roadmapping have largely focused on trend identification and content generation, the broader potential of agentic systems across all phases of roadmapping remains underexplored. This paper examines how LLM-based AI agents can support and augment roadmapping activities, and proposes a set of design considerations to guide their effective integration. Drawing on agentic design patterns, technology management principles, and recent empirical studies, we introduce a framework that outlines six core requirements for building usable, adaptable, and explainable roadmapping systems. We also highlight specific opportunities where these agents can be implemented and add value. The findings aim to inform the future development of AI-augmented roadmapping tools and contribute to the advancement of strategic foresight capabilities.

With the development of instant messaging tools, an increasing number of emojis are being used in daily online communication. Different individuals may have varying interpretations of the same emoji or other "visual symbols". These differences in understanding can sometimes lead to difficulties in comprehending the meaning of conversations. Previous research has shown that there are widespread differences in emoji usage habits across different age groups. In this paper, we focus on the differences in the understanding of various emojis between parents and children when communicating through instant messaging tools. To help parents better communicate with children, we introduce large language models (LLMs) to analyze conversations, aiming to assist parents and children in better understanding their online interactions. Through the experiment and evaluation, LLMs show the ability to be a solution for improved cross-generational communication for emoji understanding.

Delays and inefficiencies continue affecting projects across various sectors, especially large and complex projects. Recently, Artificial Intelligence (AI) has emerged as a significant asset, supporting predictive analytics, automation, and team cooperation. Utilizing these abilities, AI offers new ways to optimize resources and strengthen decision-making. This research examines the impact of AI implementation on project efficiency and delays in the United Arab Emirates (UAE), a nation experiencing swift digital transformation and promoting a bold national AI strategy. This study utilized survey data from 60 project management professionals and analysed it using descriptive statistics, correlation, and multiple regression, concludes that predictive analytics and automation are essential in minimizing delays and improving efficiency. In contrast, AI-driven collaboration tools show mixed effectiveness in the UAE's diverse, multicultural project environments. Key barriers to adoption include high costs, limited digital skills, and resistance to organizational change. The findings contribute to a clearer understanding of how AI can be practically applied in project management within the UAE, highlighting both the opportunities it presents and the challenges that remain. Based on these insights, the paper offers recommendations for policymakers and practitioners to encourage AI adoption, strengthen workforce capabilities, and ensure a balanced approach between automation and human oversight.

This study examines the substantial impact of knowledge management, encompassing knowledge creation, storage, sharing, and application, on organizational excellence in public service organizations in Palestine. The study concentrated on 220 employees at higher administrative levels, chosen by a random stratified selection technique. A questionnaire was created, the data were analyzed, and hypotheses were tested using SPSS. The findings indicated a robust positive correlation between all dimensions of knowledge management and organizational excellence. Previous literature shows a research gap on this effect in the public sector in developing countries. This paper aims to fill this knowledge gap by examining the contemporary framework of knowledge management and its influence on organizational excellence, providing tactics for decision-makers to effectively apply knowledge management approaches in the public sector to deliver high-quality services and achieve excellence, Emphasizing the importance of technology as a critical enabler of knowledge management and organizational excellence in the context of the changing public sector. The study lays the foundation for understanding how to integrate these technologies.

In this paper, which arose from a master's thesis, a methodology for planning maturity-specific engineering training projects was developed. After analyzing the increasing complexity of technical systems and the associated challenges for companies, SE was identified as a suitable approach to meet these challenges. It became apparent that the successful introduction of SE depends heavily on the company-specific planning of the qualification measures. Based on a comprehensive literature review, existing approaches to measuring maturity levels and qualification planning were examined. From this, requirements for a practice-oriented methodology were derived. The methodology developed takes into account the specific maturity level of a company, summarizes diverse role profiles in a standardized way using role clusters and evaluates the necessary competencies using a competence scale. The application in a fictitious example company showed that the methodology can be used to implement tailor-made qualification planning. In particular, the consideration of the company's maturity level and skills profiles makes it possible to design individual learning paths for different target groups. This ensures that both basic training courses and specialized further training measures can be offered in line with demand.

Given the lack of governing laws and theories specific to innovaation management noted in academic literature, the question emerges how allocation of resources by firms can be captured in terms of patterns. One possibility resides in the logic and impact of the innovation funnel. Therefore, this paper derives the law of constant innovation effort, which states that summative efforts during successive stages of the innovation funnel remain constant. This law has implications for both theory and practice in innovation management, and new product and service development in firms, sectors and (regional) networks.

Over the last few decades, knowledge management (KM) has gained a considerable attention due to management processes changes and technology progress and due to a spreading awareness of the importance of knowledge as a competitive advantage and as a key asset towards business performance enhancement. This article seeks to explore the concept of Knowledge Management within the Aviation Sector (AS). 149 documents indexed Web of Science and Scopus published between 2000 and 2023 were obtained through a Systematic Literature Review following PRISMA protocol. Bibliometric analyses were performed using R software, RStudio application and its interactive web interface called biblioshiny dedicated to quantitative science mapping analysis presented under visuals and tables. 16 relevant grey literature publications were collected and examined as well. The contribution of this study to the literature is the introduction of an overview of KM within aviation sector. The findings of this paper have indicated the increasing interest over years of the KM processes implementation particularly within the airlines industry, however it should be highlighted that the literature on this topic in aviation public sector is still not well developed especially for Air Navigation Services Provider (ANSP) organizations.

This study evaluates how sociotechnical transition frameworks can guide the adoption of human-machine interfaces and haptics in mobile machinery under the Operator 5.0 paradigm. We combine a state-of-the-art review with transparent snowballing, project-based expert evaluations, and stakeholder workshops conducted in 2025. A multi-system event-sequence analysis is used to connect ecosystem roles, platform governance, and standardization dynamics with firm-level technology related decisions. The review and expert synthesis reveal a fragmented landscape across sensing, controls, safety, and user experience. We identify four cross-cutting priorities for near-term progress: health- and state-informed assistance that leverages existing machine signals; digital-twin-in-the-loop validation to gate field trials; explainable shared control and clear authority handovers; and operator-centric metrics (workload, trust, safety-envelope adherence) as first-class requirements alongside productivity and energy. The study consolidates a taxonomy of human-machine interface and haptics for mobile machinery, maps ecosystem roles, and sets out a research agenda to bridge the lab-to-field gap and accelerate safe, human-centric deployment consistent with Operator 5.0.

The relationship between a management focused on quality and the innovation capacity of companies has raised some questions. Is there a relationship between a management focused on quality and the innovation capacity of companies? Does a business management focused on quality provide a favourable or harmful environment for innovation? This study proposes to answer the research questions raised above. The aim will be to identify similarities and differences in several organizations in order to find indications that support the answers to the questions raised. Secondary data was used from a questionnaire of a survey applied in 2020 in 257 companies. Multivariate analysis techniques were used to assess the relationship between a quality-oriented organisational culture and a favourable environment for innovation in organisations. The analysis showed that organisations ranked higher on innovation factors tend to have better rankings on quality factors, while those ranked poorly on innovation have poor quality-focused management.

The impact of technological advancements on satellite antenna tracking systems' operations and maintenance was investigated in this research. Changes in systems technology are largely driven by growing competition and the creation of innovative products. Since some systems are built in accordance with specific client requirements, customer specifications and requirements are also important elements in the advancements in earth station technology. The research method employed in this study was a quantitative method. The online survey approach was used in the data collection process. A questionnaire for satellite earth ground station staff members was developed as a research instrument and distributed to selected antenna systems maintenance, operation department, and information and communication technology employees through the company email. These are experienced employees in antenna tracking systems and subsystems for telemetry, tracking & commanding. This study's findings revealed that technological changes have improved and positively impacted the system's performance, maintainability, manageability, operations, and usability. Maintenance costs, system reliability, and efficiency are the main factors driving technological change in satellite antenna tracking systems. The study's findings show a relationship between the variables or factors investigated. This study concluded that changes in technology have positive outcomes, as the data collected from the organization and study findings show a positive impact on maintenance costs, which shows improvement. Changes in technology have reduced maintenance activities. System upgrades and new antenna tracking systems technology are recommended in this study, as technological changes in satellite antenna tracking increase efficiency and reduce maintenance expenses

The κ-η regression method, presents a significant computational challenge for hyperparameter optimization due to its inherent quadratic time complexity for training, (O(N2), which renders exhaustive grid search on large datasets pro- hibitive. This paper investigates sampling strategies as a means to mitigate this complexity. We evaluate the efficacy of performing hyperparameter tuning on a reduced subset (25%) of the training data using two distinct methods: uniform random sampling and Locality-Sensitive Hashing (LSH). Our empirical analysis, conducted across eight UCI regression datasets, demonstrates that both sampling techniques yield substantial reductions in computation time, ranging from 22% to 82%, compared to a full grid search. LSH-based sampling shows consistent per- formance in estimating the hyperparameters of the algorithm. The performance of LSH is attributed to its ability to generate a representative subsample that preserves the local geometric structure of the data, a property to which the distance-based κ-η regressor is highly sensitive. The findings validate LSH-based sampling as a robust and efficient methodology for optimizing the κ-η algorithm, offering a compelling trade-off between predictive accuracy and computational cost.

The S-curve model is widely used to describe the typical trajectory of technological performance over time, characterized by slow initial progress, a phase of rapid improvement, and eventual stagnation. However, empirical studies indicate that many development trajectories deviate from this pattern. This paper proposes a typology of six ideal-type S-curve archetypes that reflect the variety of observed development paths: the classic S-curve, early breakthrough, delayed breakthrough, steady mover, rejuvenated plateau, and non-starter. The typology is developed using a theory-building approach based on conceptual reasoning and grounded in existing literature. It provides a structured framework for analyzing technological development in contexts where quantitative data are limited or fragmented. The archetypes are intended to support early-stage technology assessment and strategic decision-making by offering a common basis for comparison across cases. Rather than predicting specific outcomes, the framework enables a systematic interpretation of development patterns and helps to identify recurring dynamics in technology evolution.

Indian classical music has been a key cornerstone of the Indian identity and its diasporas around the world. Hindustani music has always been played as vocal and instrumental music for festivities and cultural events. It plays a key role in all religious occasions. In many developing countries encompassing the Indian diaspora, there has been a decline in the number of skilled and amateur musicians of instruments such as the Harmonium, Sitar, Tabla and other classical Indian musical instruments. This trend can be easily noticed during religious and cultural events in which there is a lack of musical accompaniment for vocalists. To address this limitation, we have created a low cost tool that can analyze the melodies and rhythm in a song and generate a Tabla accompaniment to it in real time. This system was developed using Fast Fourier Transformation (FFT), onset analysis and autocorrelation to identify the {\em tempo} (BPM), {\em taal} (beat cycle length), and {\em sam} (start of song cycle) from an audio sample at the beginning of the song. A corresponding tabla accompaniment is then generated and played based on the features identified from the song. This approach only requires a mobile phone for capturing the audio and processing the data and a Bluetooth speaker for playing the Tabla accompaniment.

Time series data are necessary in many domains but analyses may be limited by availability, scale of data and privacy concerns. Synthetic time series may be used to supplement existing data to enable analyses and model building, in particular, machine learning applications. In this paper, a model for the generation of synthetic time series is proposed. This model is relatively easy to implement and aims to preserve statistical properties and temporal dynamics of the original time series. The model was found to be of higher quality to existing methods, producing synthetic time series which preserve the mean, variation and autocorrelation. Future work will examine these discrepancies and seek to preserve additional statistical properties.

Public-sector agencies are accelerating hybrid work adoption while seeking to preserve collaboration, equity, and operational efficiency. We present a collaboration-aware scheduling pipeline that combines project associations and reporting-line proximity constraints. This is achieved by clustering employees based on these associations and then mapping clusters to shared in-office days via a fairness-constrained cyclic scheduler. We introduce the Collaboration Ensurance Score (CES) to evaluate whether teams with shared dependencies are co-scheduled on the same days. Experiments on organizations of size 10, 100 and 1000 employees were ran to demonstrate the effect of problem size on performance. The Mean-Shift algorithm achieves 47% higher CES than the K-Means++ algorithm at medium scale. The proposed approach provides deployment guidance while ensuring equitable schedules and collaboration.

This paper addresses the challenge of scrap blending optimization in electric arc furnace steelmaking under uncertain scrap compositions. To mitigate the risk of violating material specifications, we develop a distributionally robust optimization (DRO) framework that combines the Wasserstein metric with a Conditional Value-at-Risk (CVaR) risk measure. Alloying constraints are expressed in terms of scenario-wise violations, and a robust CVaR bound is imposed that holds for all probability distributions within a Wasserstein ball around the empirical sample distribution. The resulting convex program minimizes expected blending costs while explicitly controlling the risk of specification violations. Numerical experiments with synthetic data show that increasing robustness through the Wasserstein radius can compensate for limited training data and ensures feasibility even for out-of-sample scenarios that deviate from the assumed distributions. These results highlight the potential of Wasserstein-CVaR DRO as a tractable and effective tool for robust process optimization in steelmaking.

The security of e-learning platforms is an essential pillar for protecting sensitive user data and ensuring the sustainability of digital education systems, particularly in Africa. In our study, we assessed the security of 40 learning management systems (LMSs) from 18 African countries. For vulnerability detection and compliance analysis, we used advanced tools such as SSL Labs, Nessus, HostedScan, SOVY, internet Secure Scan and the Mozilla HTTP Observatory. E-learning platforms were analyzed by region and typology (governmental, private, academic), highlighting significant disparities. The results revealed critical vulnerabilities, such as remote code execution flaws and obsolete SSL/TLS configurations, as well as shortcomings linked to players' lack of knowledge of standards, limited budgets and low prioritization of security at the design stage, to the detriment of good functional coverage. To meet these challenges, we propose the OSSEP (open-source security for educational platforms) framework, a model created and adapted to the African context, that integrates open-source tools and recommendations aligned with international standards and security practices. These practices are aligned with LMS security requirements, including those using artificial intelligence (AI). Qualitative feedback obtained via a focus group of four IT security experts confirmed the relevance of OSSEP while calling for practical, longitudinal experimentation to validate its effectiveness in real-life contexts. Our study highlights the vulnerabilities of LMSs in Africa and proposes a structured, cost-effective solution for improving their security, thereby helping to increase confidence in digital education on the continent.

In the rapidly evolving digital economy, sustainable digital transformation in public-sector organizations is increasingly dependent not only on technological adoption but also on cultural alignment. This paper presents a longitudinal case study of a legacy public-sector bank in India, referred to as United Bank of India (UBI), and explores how digital culture has been strategically cultivated and aligned to enable and sustain digital transformation. Using the Assess-Align-Adapt-Activate (AAAA) framework, this study examines the transformation journey over 2.5 years, identifying how values like innovation, agility, customer-centricity, and digital literacy have been embedded and evolved. Through interviews, observations, and document analysis, we provide a process-oriented understanding of culture as an enabler, not just a context, for transformation. The findings contribute to the theory of digital change by positioning culture as a dynamic infrastructure and offer a roadmap for policymakers and practitioners seeking to drive inclusive and sustainable transformation.

In the era of rapid technological advancement, Digital Engineering Transformation (DET) has become imperative for organizations seeking to maintain competitive advantage. This paper introduces a comprehensive framework that delineates the fields of actions in a Digital Engineering Transformation. Anchored in the strategic direction of the enterprise or product, the framework identifies four critical fields of action: product, processes & methods, IT-architecture, and organization. The activities on the product level focuses on the advancement and integration of digital development technologies which enhance the characteristics of the product directly. Processes and methods aim to optimize efficiency and innovation. The IT-architecture underpins scalability and flexibility, while the field of the organization emphasizes cultural and structural adaptation. Drawing on established theories and real-world case studies, this framework provides a structured approach to understand and implement a Digital Engineering Transformation. This paper aims to serve as a foundational reference for academics and practitioners navigating the complexities of Digital Engineering Transformation.

Surgical care requires real-time coordination among patients, clinicians, devices, and data, often under unpredictable and time-critical conditions. Yet, current hospital systems struggle with adaptability and integration, limiting their ability to respond dynamically during surgery. Digital Twin (DT) technology, initially developed in aerospace and manufacturing, is now emerging as a promising tool for enhancing situational awareness and supporting real-time clinical decisions. This paper presents a multi-layered DT Decision Support System (DT-DSS) architecture tailored to the surgical domain. Drawing from recent literature and scoping reviews, we identify current limitations in model fidelity, system integration, and ethical transparency. We propose a modular framework that spans data ingestion, simulation, predictive optimization, clinical decision-making, and governance. Each architectural layer is backed by validated studies, offering a practical and trustworthy pathway toward real-world surgical implementation.

While Digital Transformation (DT) in railways promises significant benefits, it also introduces substantial complexity across technical, organizational, and institutional domains. These challenges must be recognized and addressed from the front-end phase of DT projects, where early risk identification is critical to project success. In response to the absence of a framework tailored to the unique characteristics of railway DT initiatives, this paper introduces D-TRACK (Digital Transformation RAilway Complexity frameworK) -- a structured tool designed to support the early identification of complexity-related risks in railway DT projects. Building on the UK National Audit Office (NAO) Delivery Environment Complexity Analytic (DECA) framework, D-TRACK integrates domain-specific sources and practitioner insights to reflect the distinctive realities of the railway sector. The framework comprises twelve complexity factors organized into four overarching dimensions, each supported by guiding questions that enable a structured and systematic risk identification process. To validate D-TRACK, a serious game (SG)-style workshop was conducted with eleven experts in digitalization and railway operations. The workshop confirmed that D-TRACK effectively met its three core design objectives: railway relevance, digitalization compatibility, and risk identification support. In addition, participant feedback underscored the broader value of SGs -- not only as validation tools but also as enablers of team alignment during the formative stages of complex projects.

As cyber threats increasingly target human vulnerabilities, understanding behavioural and organisational factors is vital for strengthening resilience. It is essential to evaluate cybersecurity awareness among public sector employees, and this study has used Saudi Arabia as a case study with a specific focus on their vulnerability to social engineering attacks. The literature review examines key behavioural theories, including Protection Motivation Theory (PMT) and the Theory of Planned Behaviour (TPB), to investigate how motivation and intention influence cybersecurity practices. Established frameworks such as the NIST Security Awareness, Training, and Education (SATE) model, the Human Aspects of Information Security Questionnaire (HAISQ), and the Cybersecurity Awareness Training (CSAT) framework are critically assessed for their effectiveness and limitations, especially in measuring training outcomes and adapting to diverse organisational contexts. The review also categorises key social engineering threats, including phishing, spear phishing, pretexting, baiting, tailgating, and quid pro quo, highlighting their mechanisms and mitigation strategies. Using a quantitative, positivist methodology, the study employs online surveys analysed via SPSS. Convenience sampling and a cross-sectional design support broad data collection while maintaining ethical standards. The result was presented statistically, indicating that cybersecurity training, organisational policy, and culture significantly enhance awareness. The findings underscore the need for continuous, tailored training programs and policy reinforcement. Future research should adopt longitudinal and mixed methods approaches, expand sample sizes, and explore emerging technologies, such as artificial intelligence, to develop adaptive cybersecurity awareness strategies tailored to the evolving demands of public service.

In recent years, there has been a massive interest in Artificial Intelligence (AI) Adoption, and AI applications have been introduced in various industries. AI adoption refers to how organizations, industries, and governments integrate AI technologies into their operations, products, and services. Adopting AI can improve many aspects of business, including supply chain management, customer service, predictive analytics, and repetitive work automation. However, despite the possible advantages, adopting AI comes with obstacles businesses must overcome to guarantee a successful rollout. Therefore, this research aims to (1) identify the factors impacting AI adoption decisions and (2) develop a model to evaluate AI adoption challenges using a hierarchal decision model (HDM) to determine the ranking of the impact of these factors and increase the success rate of AI adoption.

The increasing decentralization and digitalization of energy distribution systems pose significant challenges to the management of operational and maintenance processes. The integration of renewable energy sources and the resulting dynamic grid conditions necessitate more adaptive, efficient, and technologically advanced workforce management (WFM) solutions. This work presents a systematic approach to modeling WFM processes tailored to the needs of modern smart grids. Based on a detailed evaluation of modeling languages and a requirements analysis informed by industry workshops, Business Process Model and Notation (BPMN) was identified as the most suitable formalism. An iterative development process was established, combining user story specification, process modeling, validation, and requirement refinement. The resulting WFM processes integrate real-time reactivity and AI-based decision support mechanisms. These include intelligent task classification, personnel selection, and resource planning, enabling human operators to make faster, more consistent decisions in dynamic grid environments. The work lays the foundation for a digitalized, flexible WFM-System that addresses the future demands of sustainable and secure energy distribution

Smallholder farmers in Northern Thailand often rely on personal experience and short-term weather data to make time-sensitive crop planning decisions, such as flowering induction or irrigation. However, variability in growing conditions and uncertain production timing hinder reliable outcomes. This study proposes a Forecast-Based Decision Support System (FDSS) to help farmers anticipate key actions and organize workflows more systematically. Using the DMADV (Define, Measure, Analyze, Design, Verify) framework, the research focuses on developing a modular system grounded in fieldwork. Semi-structured interviews with young longan farmers identified operational pain points, informing five core modules: (1) Data Ingestion, (2) Predictive Analytics, (3) Visualization, (4) Recommendation Engine, and (5) Knowledge Exchange. The system emphasizes low-barrier adoption, local relevance, and alignment with existing practices. Feasibility was assessed through complementary analyses, including SWOT, TOWS, and TAM-SAM-SOM frameworks, demonstrating strong potential for scalable deployment within the targeted smallholder context. This paper suggests how a user centered FDSS can transform fragmented data into timely, actionable insights. By aligning forecasting tools with farmers' decision cycles, the system enhances planning without compromising autonomy. The findings offer design implications for modular, accessible decision support tools in small-scale agriculture.

Project management tailoring is fundamental to the success of development projects, but requires a deep understanding of the complex factors that influence a project and its management. This paper presents INFACTS, a comprehensive structuring model that contains project-, organization-, product- and person-oriented influencing factors that have an impact on the selection and tailoring of project management. INFACTS integrates a theoretical knowledge base with a practical decision- making aid for project management tailoring. By combining literature analysis, terminology work, and qualitative content analysis, it translates complex project situations into a structured visual framework. Each influencing factor was assigned five qualitative or quantitative characteristics, enabling a clear positioning between the spectrum of plan-based and agile project management. A real-world R&D project case study demonstrated the applicability and value of INFACTS. It showed that early context analysis increases transparency and can lead to more informed decisions during tailoring, thus proactively preventing misalignments in the project management approach. INFACTS can be used independently of a specific project management method, closes a key research gap and lays the foundation for further research in the field of modern project management.

E-commerce's rapid increase in emerging nations, mainly Brazil, has brought with it operational obstacles and opportunities. Multiple merchants and customers are combined into structures like Olist, which increases the likelihood of delivery delays, chargeback fraud, and unsatisfactory customer experiences. The openly accessible Olist dataset was used in this study. This paper proposes a learning framework for operational chance prediction. We created a facts pipeline that combines complex models like LightGBM and XGBoost with preprocessing, feature engineering, and controlling class imbalance. Crucial risk drivers, along with freight-to-price ratios, installment systems, and delivery delays, were identified using SHAP-based explainability so as to improve interpretability. With an F1-score of 0.81, our results show that LightGBM outperformed random forests and logistic regression in terms of predictive functionality. A decision-support system that converts risk scores into practical mitigation strategies, such as expedited cargo, fraud detection, and customer recovery initiatives, is another idea we include. In the end, simulation analyses indicate that focused interventions can decrease high-risk instances by more than 25%. This study helps e-commerce operational risk management by integrating selection help and predictive analytics, offering platform and logistics partners beneficial records.

The primary contribution of this paper lies in evaluating the potential benefits of using machine learning algorithms for channel selection for wireless communication systems on the 5GHz spectrum. In this paper we explore classification algorithms such as Decision Tree (DT), Random Forest (RF) and Neural Network (NN) with multiple outcomes by considering multiple decision factors. This paper demonstrates, though simulation scenarios, the effectiveness of classification algorithms, and the results show that this approach can provide extra stability for the wireless communication network with the scenarios considered.

Innovative technologies, such as smart homes, are considered instrumental in contributing to environmental sustainability. However, despite the anticipated benefits, the adoption of smart home technologies remains limited. This necessitates exploring the match between the expected and actual realization of smart home sustainability potential, which has not received sufficient attention. To address this gap, this study adopts Expectation-Confirmation Theory to explore the (mis)alignment of smart home performance with expected sustainability benefits and the conditions contributing to it. Drawing on survey data from 244 UK-based smart home users, the findings shows that while many users initially adopt smart home technologies to achieve energy efficiency, their actual usage behavior often falls short of this goal. There is a limited engagement with sustainability measurement tools because of barriers, namely, lack of time, experience, and equipment, the complexity of sustainability measurement, and prioritization of financial savings over sustainability. Such findings advance theory by offering insights into the behavioral factors that shape technology performance perception in the context of sustainable technology use, which also offer implications for policymakers.

Traffic Generator (TG) aims to create a virtual traffic world for many different systems, such as the Driving Simulator (DS) or Motorbike Simulator (MS). Especially, TG can be utilized as a Traffic Accident Simulator (TAS) by inputting many scenario parameters (physics, vehicle location, etc.). For instance, the Police can use TAS to analyze and report traffic crash results to the Procuracy for review and decision-making. Furthermore, TG can be "zoomed out" by converting to a Mesoscopic model for a Traffic Network Management Simulator (TNMS). Due to the multiple purposes for TG, the process of TG development is varied. Hence, the quality assessment standard for the TG development cycle needs to combine multiple disciplines that must be related to the objectives of the TG project type. Nevertheless, current standards like CMMI and ISO cover either project management, traffic engineering or software engineering or other fields. This separation creates some challenges that only emerge when combining two or more disciplines. For instance, TAS requires realistic details to simulate the sequence of traffic accidents between vehicles and infrastructure, which demands a combination of not only traffic, software, and project management but also architecture and construction. Thus, this study proposes the Traffic Information Technology standard for developing the Traffic Generator project by utilizing our previous study of the 6R process, the APAS formula and the idea of TG-verse. In this research scope, we contribute the challenge list that our TIT standard can solve based on our perspectives, knowledge and experience in traffic, management, software, law and architecture.

Autonomous Vehicles (AVs) are becoming central to the mobility infrastructure of smart cities, promising safer, cleaner, and more efficient transportation. Yet, their adoption also introduces significant ethical, social, and regulatory challenges that extend far beyond technical deployment. This paper presents a multi-dimensional framework grounded in Rawlsian justice theory and Value-Sensitive Design to evaluate and guide the ethical integration of AVs into urban ecosystems. Drawing on empirical evidence and simulation using the Cityscapes dataset, we identify key areas of concern including perception bias, spatial inequity in AV deployment, data privacy risks, and surveillance implications. Our findings reveal that AV systems, if unregulated, may exacerbate digital redlining and reinforce socio-economic disparities. We propose ethical risk modeling and decentralized governance mechanisms to mitigate these risks and promote transparency, inclusivity, and sustainability. By combining theoretical ethics with technical simulation and policy analysis, this study offers a comprehensive roadmap for cities aiming to align AV innovation with principles of justice, trust, and resilience.

This study presents a comprehensive review of recent research addressing routing, scheduling, and operational challenges in autonomous electric vehicles and unmanned aerial vehicles within logistics and mobility systems. Unlike previous surveys that provided broad overviews of vehicle routing problem variants and electric vehicle sharing systems, our review focuses specifically on the integration of autonomy and electrification, with particular attention to collaborative ground and air delivery and shared autonomous mobility. The surveyed works highlight three main directions: the incorporation of realistic energy-related constraints such as nonlinear battery consumption and partial recharging; the diversification of applications ranging from last-mile delivery and warehouse stocktaking to ride-sharing and large-scale shared mobility; and the methodological variety spanning exact optimization, heuristics, metaheuristics, and hybrid approaches. These advances mark a shift toward models that better balance realism and efficiency in autonomous logistics. As a result, the study outlines several promising directions for future research in this area.

The increasing penetration of renewable energy and the growing demand for cross-border electricity transfer through interconnection pose significant challenges to voltage stability in modern smart grids operation. Among the various Flexible AC Transmission Systems (FACTS) devices in power system management, the Static Synchronous Compensator (STATCOM) has emerged as an effective solution for dynamic voltage regulation and reactive power control. This paper presents a review of STATCOM applications in smart grids operation, focusing on its role in maintaining voltage stability and enhancing power flow. To complement the review, a case study is developed on a 500 kV, 150 km British Columbia (Canada) and Washington (USA) grid interconnection line. A MATLAB/Simulink model is implemented to evaluate system behavior under normal operation and fault conditions, both with and without STATCOM. Simulation results demonstrate that STATCOM significantly improves voltage recovery, reduces oscillations, provides rapid reactive power support, and stabilizes the DC link voltage. These findings confirm the practical importance of STATCOM in smart grid operations, particularly in long-distance and high-voltage interconnections. Future directions are also highlighted, including advanced control strategies and renewable energy integration in Energy Management Systems of Smart Cities.

The rapid development of Generative Artificial Intelligence (Gen AI) based on foundation models (GPT-4, DALL·E, and Claude) is capable of altering entrepreneurial ecosystems with evolving content creation, predictive analytics, and dynamic mentoring. Although the Gen AI application is rapidly expanding in all sectors, its practical absorption in the context of technology-orientated business incubators (TBIs) is an insufficiently developed area. In this study, this gap is covered by using empirical research to explore the impact of Gen AI on improving one in the quality of startup selection, improvement of innovation capability, and shortening of time-to-market in incubation settings. On the basis of the Technology-Organisation-Environment (TOE) framework and Entrepreneurial Ecosystem Theory, a quantitative research design will be used by referencing data provided by the incubator managers and the founders of the emerging enterprises, which operate in various industries. With the partial least square structural equation modelling (PLS-SEM) and then following the said methodological approach with robustness checks done using the two-stage least squared (2SLS) regression, the results lend credence to the significant positive effects of Gen AI adoption on the startup performance outcomes. It is worth noting that AI-assisted mentoring proved the most decisive change, implying the possibility of scalable, individualised capability building. Findings provide both theoretical implications for the application of Gen AI capabilities in relation to stages of the incubation lifecycle and practical implications for incubators, entrepreneurs, and policymakers. This study establishes Gen AI as a business strength in terms of promoting inclusive, sustainable and innovation-based entrepreneurial ecosystems.

The increasing adoption of Large Language Models (LLM) and Retrieval-Augmented Generation (RAG) pipelines is transforming information access, yet their efficacy is significantly hampered in developing regions by a scarcity of relevant, localized data and compute. Furthermore, reliance on centralized RAG architectures raises critical data governance concerns, as organizations and governments are reluctant to forfeit control of sensitive data, thereby centralizing economic benefits and exacerbating regional digital divides. To address these challenges, this paper introduces a novel framework that leverages a distributed network of specialized LLMs rooted in the principles of Distributed Retrieval-Augmented Generation (DRAG). Our proposed architecture eliminates the need for a centralized knowledge base, allowing data owners to retain full control over their assets. We detail an intelligent query-routing mechanism for efficient knowledge discovery within the network and, critically, propose a comprehensive incentive model to encourage the active participation of regional entities. By fostering a collaborative, yet decentralized, ecosystem, this framework facilitates equitable data governance and paves the way for a sustainable, regional data economy, empowering local communities and enhancing the accuracy of LLMs.

Software requirements are typically written in natural language, which is prone to issues such as ambiguity, wordiness, and vagueness. These requirements issues cause faults in software development processes. In the past, a well-known EARS (easy approach to requirements syntax) templates have been proposed to mitigate these issues by specifying the requirements in semiformal format. The requirements specification in the EARS format requires an understanding of the EARS and requirement engineering domain. This requires manual labor, where each requirement needs to be manually transformed, looked upon, rewritten, and transcribed by a requirement engineer. We propose a novel framework for fine-tuning light-weight LLMs that can automatically rewrite the requirements from scratch. We designed and implemented a pipeline that transforms the raw requirements to EARS format by fine-tuning four LLMs, that is, BART, DistillGPT2, GPT2, T5-Small with BART using data set of 9141 raw requirements. The BART-base provides the highest average BLEU score (0.405) and the lowest final training loss, indicating strong performance in transforming the EARS requirement.

The rapid evolution of project complexity in today's business environment demands more sophisticated control mechanisms than traditional project management can provide. This paper proposes a framework that harnesses artificial intelligence to transform the monitoring and controlling of the projects from inception to completion. Unlike conventional approaches that depend heavily on periodic reports and static analysis, the proposed system integrates machine learning algorithms, natural language processing, and predictive analytics to create a dynamic, responsive project control environment. Through careful examination of existing literature and practical observations from real-world implementations, the paper has developed a five-module framework that addresses critical gaps in current project management practices. The proposed model monitors and learns, predicts, and adapts so that offers project managers unprecedented visibility into potential issues before they become problems. While acknowledging the inherent challenges of implementing AI in complex organizational structures, this work provides a practical roadmap for organizations seeking to modernize their project control capabilities.

Over the past decade, the convergence of artificial intelligence (AI) and customer experience (CX) has reshaped business strategy, customer engagement, and digital transformation. This study presents a comprehensive bibliometric analysis aimed at mapping the scientific landscape of AI and CX studies published between 2015 and 2025. Using data extracted from the main Web of Science collection and analyzed via VOSviewer and Biblioshiny (R), the study applies performance analysis, co-citation, keyword co-occurrence, and thematic mapping to highlight publication trends, key contributors, influential journals, and emerging research themes. The results show a considerable increase in academic output since 2018, with a compound annual growth rate of 78.62%, driven by growing interest in AI applications in customer-centric strategies. Thematic analyses highlight the relevance of concepts such as personalization, chatbots, and predictive analytics, while revealing conceptual fragmentation and geographical imbalances in research contributions. Furthermore, the study highlights significant gaps in ethics, contextual diversity, and interdisciplinary integration. These insights provide a structured foundation for future research and guide academics, professionals, and policymakers who wish to leverage AI for groundbreaking CX strategies.

This research study was aimed at investigating factors contributing to mobile crane breakdowns and assessing the alignment of the case company's maintenance practices with Total Productive Maintenance (TPM) best practices. The research study was conducted at a company that owns, operates, and maintains a fleet of cranes using a single-case, mixed-methods approach. Maintenance data from the company's database was analysed, and surveys were administered to employees involved in crane maintenance and operation to gain deeper insights into the causes of mobile crane breakdowns and maintenance practices. The findings revealed that multiple factors, such as the age of the cranes, operator errors, and negligence, contributed to mobile crane breakdowns. Furthermore, the findings revealed that the case company's maintenance practices are misaligned with TPM's best practices in areas such as planned maintenance, autonomous maintenance, focused improvement, and early equipment management. The study provides recommendations for closing the identified gaps and optimising maintenance practices in line with TPM principles

The Integrated Functional Model framework of a Glassdoor Refrigerator in a Smart Home setting is presented in this paper. Principles of Prototyping, Conceptual Design, Concurrent Engineering, Quality Function Deployment, House of Quality, Voice of Customer, Analytic Network Process, Design Structure Matrix, and Kaizen have been implemented in a novel redesigning technique to achieve promising results.

In the rapidly evolving landscape of Industry 4.0, predictive maintenance is crucial for maximizing operational uptime and efficiency. Traditional predictive maintenance approaches, often reliant on static models, face challenges in addressing novel anomalies and providing contextually rich, actionable insights. An innovative Retrieval-Augmented Generation (RAG) system is proposed, leveraging Large Language Models (LLMs) for real-time anomaly detection and predictive maintenance. The system dynamically integrates machine sensor data with repair manual documentation and worker insights to generate highly relevant and interpretable diagnostic maintenance recommendations which are passed along in real-time to relevant personnel. The methodology aims to demonstrate the RAG system's ability to enhance decision-making, optimize industrial operations and mitigate equipment failures in complex manufacturing environments.

In the last decade, intensified competition, global risks, and reliance on competitors' pricing strategies in the truck and bus industry have compelled companies to monitor and track their costs with increased precision. Stringent cost monitoring, calculation, and budget management are essential to ensure sustainable growth and long-term profitability, thereby elevating the significance of the cost management department. Additionally, companies strive to shorten the time-to-market to outpace competitors and maximize early market opportunities. The rapidly changing environment necessitates that cost management adopt shortened, iterative, and renovated approaches. In recent years, various ML-based models have been developed to address these challenges. Among other machine learning models, support vector machine (SVM) models come forward for parametric cost estimation because of their robustness to noise and outliers, efficacy with non-linear relationships, and good success rate with small datasets. Rotational molding has become an increasingly preferred manufacturing method due to the growing demand for low-volume plastic parts. This paper focuses primarily on the development of an improved SVM model with various ensemble methods that capture the cost estimation relationship between the characteristics of rotational molded parts and rotational molds. The proposed ML-based model achieved 78.4% accuracy (MAPE = 21.6%), reaching the predefined targets. Hence, the developed model was offered for the use of cost engineers via the development of a user-friendly graphical user interface (GUI), which allows them to benefit from the model in the early development phases.

We constructed a Retrieval Augmented Generation (RAG) system for industrial machine manuals to quantify the performance of various open-source Large Language Models (LLMs). Using technical manuals from a flour mill, textual data was extracted using MinerU. The data was processed using Natural Language Processing (NLP), embedded with "Qwen3- Embedding-0.6B" and entered into a PostgreSQL database using LangChain's pgVector API. Our RAG pipeline integrated a tag extractor, three query expansion techniques (Multiquery, Decomposition, Step back) and one scoring algorithm (BM25), a PostgreSQL Database, a Reranker (mxbai-rerank-xsmall-v1), a Prompt template and an LLM. We evaluated the BERTScore performance of five open-source LLMs (Gemma3:1B, Gemma3:4B, Llama3.1:8B, Llama3.2:1B, Llama3.2:3B) in this RAG pipeline. One key finding is that model performance is task-dependent: Llama3.1:8b excelled in precision, while Gemma3:4b excelled in recall. Smaller models can outperform larger models in specific tasks, as demonstrated by Llama3.2:3B superior performance in identifying unanswerable questions. This study provides a framework for selecting cost-effective and high-performing open-source LLMs in industrial applications and demonstrates the effectiveness of smaller parameter models, a critical finding for compute limited businesses.

In embedded systems manufacturing, ensuring the reliability and performance of products through rigorous testing is critical. However, structural testing, particularly the auto-generation of path-focused test cases, is a complex and resource-intensive task. Traditional methods often struggle to efficiently produce optimized test cases that cover all possible execution paths. This study addresses the challenge of automating test case generation by integrating Generative AI with an enhanced giant armadillo optimization (EGAO) algorithm which efficiently generates and validates optimal test cases for embedded systems, enhancing the testing process's speed and accuracy. The methodology combines the power of Generative AI to formulate the best test case combinations with the optimization capabilities of the EGAO algorithm, which is used to search for global solutions by navigating complex solution spaces. The results of the proposed model are validated through practical applications in embedded systems manufacturing, shown improvement in test coverage, efficiency, and ability to identify critical system vulnerabilities. Experimental findings show that the proposed GAI+EGAO model outperforms traditional test case generation models, reducing testing time and increasing the accuracy of fault detection in the manufacturing process.

This qualitative study explores the challenges and practitioner needs in designing resilient automotive system architectures, addressing the growing complexity and vulnerability of connected and automated vehicles. Based on a semi-structured qualitative interview with experts across systems engineering, safety, and cybersecurity domains, the study identifies key challenges and expectations for implementing resilience-by-design in industrial practice. The findings reveal five critical themes: (1) the lack of standardized process models for the consideration of resilience in the early design phase, (2) limited methods for disruption identification using the systems architecture during early design phase, (3) insufficient evaluation metrics and methodology, like a maturity model, and (4) the absence of integrated toolchains to support resilience measures. These insights are synthesized into four major action fields that define the strategic priorities for enabling resilience-by-design: (1) structured development processes, (2) disruption analysis and resilience evaluation methods, and (3) AI- and model-based design tools.

The study aims to synthesize recent studies on the key barriers impeding the widespread adoption of electric vehicles (EVs), with a particular emphasis on the United Arab Emirates (UAE) and the United Kingdom (UK). Transitioning to EVs is crucial for mitigating greenhouse gas emissions and achieving sustainable transportation goals. However, several obstacles hinder their large-scale implementation across various domains, including technological limitations, infrastructural deficiencies, economic constraints, social and psychological factors, and policy and regulatory challenges. The research explores the similarities and differences in these barriers between the UAE and UK, two regions with distinct contexts. Technological barriers related to battery performance and charging infrastructure, along with high upfront costs and insufficient economic incentives, are common challenges. Infrastructural barriers, such as the uneven distribution of charging stations, exacerbate range anxiety. Cultural preferences and social attitudes, particularly in the UAE's affinity for luxury vehicles, also impede adoption. Psychological barriers like range anxiety and misconceptions about reliability further hinder consumer acceptance. The study highlights the pivotal role of government policies and incentives in addressing these barriers. While both regions have implemented supportive measures, the consistency and comprehensiveness of these policies vary, impacting their effectiveness. By critically evaluating the existing literature, this review aims to identify key gaps and provide insights for policymakers, industry stakeholders, and researchers to develop strategies that accelerate EV adoption, ultimately contributing to sustainable transportation.

Integrating appropriate project management practices in adopting advanced agricultural technologies is vital for enhancing sustainable growth in the UK's agricultural sector. This study examines the role of project management practices in promoting the adoption of advanced technologies in the UK's agricultural sector, with a focus on enhancing productivity and sustainability. However, this agrarian sector faces significant challenges, including stagnating productivity and increasing pressure to adopt sustainable practices. Technologies such as precision farming, IoT applications, and AI-driven decision support systems offer potential solutions to these challenges. Nevertheless, adoption rates have been uneven, primarily due to high initial costs, a lack of technical knowledge, and integration difficulties. This study reviews existing literature on technology adoption and project management in agriculture to identify effective strategies, barriers, and best practices relevant to the UK context. The research employs a mixed-methods approach, incorporating case studies and integrating quantitative data collection through structured surveys with secondary data analysis. The methodology aims to comprehensively understand technology adoption rates, project management practice effectiveness, and the impact of UK-specific policies. The study's main findings indicate that effective project management practices, particularly those based on Agile and Lean methodologies, can significantly enhance the adoption of advanced agricultural technologies. These practices address key barriers like financial constraints and technical knowledge gaps. The study suggests that improved stakeholder engagement, continuous improvement practices, and supportive policies are crucial for successful technology integration. Implications for engineering practice include the need for tailored project management training and the development of collaborative platforms for knowledge sharing. Future research could focus on longitudinal studies to track the long-term impacts of these practices and explore the adoption of emerging technologies, such as blockchain and advanced data analytics in agriculture.

Financial inclusion is a crucial issue in Morocco, particularly in the digital age. Although progress has been made, disparities persist between urban and rural areas and across socioeconomic groups. This study explores the determinants of financial inclusion in Morocco, focusing on the impact of digital financial services such as mobile money, mobile banking, and mobile experience. Data analysis highlights how these technologies expand access to financial services, particularly for vulnerable populations. However, challenges remain, notably regarding perceptions related to the costs of financial services. The results show that greater integration of digital technology into the financial sector could significantly improve financial inclusion in underserved regions. This research offers recommendations to strengthen the impact of digital technologies and promote more equitable financial inclusion.

Haptic feedback is crucial for enhancing realism and immersion in XR applications, yet current implementations often lack consistency across devices and interaction types. In this paper, we propose a universal approach for designing real-time haptic feedback in the Meta Quest ecosystem, based on a modular architecture and an interaction-driven rule. Our key insight is that the timing of haptic feedback should depend on the semantics of interaction: feedback should be triggered upon physical contact if the object does not change its state, and upon state change otherwise. We implemented this principle in Unity using a layered architecture that decouples hand logic, application logic, and glove-specific hardware control. The framework supports both custom-built (Haptic CeTI Glove) and commercial (TactGlove DK2) gloves and has been validated in two XR applications: a juggling simulator and a latency testing interface. Our results demonstrate how architecture-level abstraction enables reusability and extensibility across use cases, while user testing highlights the importance of latency in perceived realism. The proposed rule provides a practical guideline for developers aiming to design intuitive and consistent haptic experiences in XR.

Generative AI and Financial Technologies are the two disruptive technologies that are transforming the digital landscape into different domains by enhancing their efficiency such as financial services, banking, insurance, risk assessment, investment, wealth management and payment methods. LLMs and generative AI have introduced radical shifts in the fintech industry. In fact, the mix between these two technologies holds tremendous potential to create new business models enabled through digitalization. Although research on GenAI & Fintech Applications has emerged, understanding the utility of its integration for business remains limited. This study aims mainly to characterize the benefits of integrated AI and Fintech across different finance applications. Using bibliometric analysis, this study highlights the most influential articles on the subject based on their publications, citations, and importance in the Research Community network.

Kenya's mobile-financial ecosystem, driven by M-Pesa and a mobile penetration rate above 133%, has advanced commerce and financial inclusion but also exposed users to SMS-based scams exploiting linguistic diversity, cultural trust, and psychological manipulation. Existing fraud-reporting mechanisms such as keyword filtering and manual reporting, remain reactive and inadequate against these evolving threats. To fill this gap, this study develops a culturally adapted, machine learning-driven approach to SMS scam detection tailored to Kenya's multilingual environment. Using a crowdsourced dataset of 738 SMS messages (427 scams, 311 legitimate), XGBoost achieved 83.8% accuracy with strong precision and F1-scores, while Logistic Regression offered superior recall (91.8%) in detecting fraudulent content. SHAP analysis revealed urgency cues, code-switching features, and social proof language as the most influential predictors of scams. The proposed model, designed for integration via APIs into USSD platforms, enables real-time detection and fosters user confidence through explainable alerts. This study provides a scalable, context-aware framework for fraud prevention in mobile-first economies and actionable insights for telecom operators, fintech platforms, and policymakers seeking to strengthen digital trust.

Accurate and granular sales forecasting becomes a strategic necessity in the dynamic realm of intelligent retail, and this paper puts forth a cutting-edge hybrid framework fusing together ensemble learning (XGBoost) and a sophisticated time series modeling technique (Facebook Prophet) to forecast daily-level, state-wise drugstore sales across Germany. A real-world dataset from the Rossmann pharmacy chain is employed, incorporating also internal (promotions, assortment mix, store format) and external variables (public holidays, competitors' proximity) to model sales dynamics. The XGBoost model turned in excellent predictive performance (R² = 0.85) in capturing highly complex nonlinear interactions, while the Prophet model undertook the robust trend-seasonality decomposition task using temporally varying data alone. Therefore, unlike previous models that have been limited to store- or product-specific forecasts, our approach delivers geo-aware, multi-store forecasts to gain insights on region-specific sales drivers. These findings will arm retailers with interpretable, accurate forecasts for optimizing promotional strategies, inventory allocation, and store network planning. The research presents a scalable data-driven forecasting paradigm that can be applied in high-dimensional and competitive retail environments.

The accurate estimation of fruit shelf life is crucial for maintaining product quality and reducing food wastage. In recent years, advancements in deep learning and computer vision techniques have opened up new possibilities for non-invasive and efficient shelf-life prediction. This research aims to leverage deep learning and object detection methods to predict the shelf life of fruits. In the realm of perishable goods, such as fresh fruits, optimizing inventory management and reducing waste is an ongoing challenge. One critical factor that significantly influences consumer decisions in this context is the remaining shelf life of the product. This study aims to present a compelling justification for focusing exclusively on the variable of remaining shelf life to predict fresh fruit purchases. This study proposes a novel approach that combines deep learning and object detection techniques to predict fruit shelf life. The proposed framework involves two main stages: fruit detection and shelf-life prediction. In the fruit detection stage, state-of-the-art object detection models are employed to identify and track individual fruits within images. Subsequently, the detected fruit images are fed into a deep-learning model designed to predict shelf life. The model takes into account various visual features such as colour changes, texture variations, and morphology, which are indicative of fruit ripening and deterioration. The integration of object detection and deep learning enhances the accuracy of predictions by capturing spatial and temporal information, enabling a comprehensive understanding of fruit conditions.

Traceability across system views is essential for enabling automation, impact analysis, and design consistency for data driven model-based systems engineering (MBSE). However, establishing trace links-especially between heterogeneous artifacts like requirements and physical components-remains a manual, error-prone task. This paper proposes a multi-feature graph learning approach for automated trace link prediction across RFLP (Requirements, Functional, Logical, Physical) views. By fusing semantic embeddings, numerical attributes, and geometric descriptors into a unified representation, we train two graph neural network (GNN) variants to predict intra-view (P - P) and cross-view (R - P) links. Experiments on a real-world demonstrator reveal that GNNs outperform classical baselines, particularly in cross-view settings, and highlight the importance of multimodal fusion and structural reasoning. Our results aim to support data-driven trace link prediction to close digital thread gaps and support downstream AI applications in engineering.

Historically, the Vector Quantization (VQ) image compression algorithm was designed for single-core processors. Despite its simplicity, impressive bit rates, and good reconstructed image quality, the VQ algorithm is limited by a runtime complexity of O(N) where N is the size of the main codebook. As each input pixel block can be processed independently, the traditional VQ algorithm does not fully exploit modern multi-core architectures. We propose novel GPU-accelerated implementation for both the VQ encoder and decoder that exploits multi-core architectures. We present the design of CUDA kernels for distributing the computations across multiple GPU threads. Although these kernels do not fundamentally change the theoretical runtime complexity of encoder and decoder, it does reduce their respective constant factors which yields significantly lower execution times. Specifically the VQ encoder is improved by 61x, 44x, 34x and 29x for main codebook sizes N = 128, 256, 512 and 1024 while the GPU accelerated VQ decoder is at least 200x faster across all codebook sizes. These improvements make VQ algorithms more practical for real-time applications such as multi-media streaming and reversible data hiding.

This paper examines the factors that shape technological innovation efficiency in 137 high-tech SMEs in Shandong Province, China. A two-stage empirical strategy is adopted. In the first stage, a stochastic frontier method is applied to estimate each firm's innovation performance. In the second stage, regression analysis evaluates the effects of corporate social responsibility (CSR), top management support (TMS), market competition (MC), and social cooperation networks (SCN). The results indicate that CSR, TMS, and MC have a significant positive impact, while SCN shows no statistically significant effect. Robustness is tested through matching techniques to mitigate selection bias and quantile-based analysis. The findings provide managerial guidance for enhancing innovation capabilities in SMEs under resource constraints.

We provide an optimization model for optical lens inventory management. This model aims to reduce costs by predicting which lenses should be pre-ordered in bulk while ordering less popular lenses through a third party which is more costly. The model takes into account prediction errors. If too many lenses of a certain type are pre-ordered then there is an associated cost because funds were invested. If too few lenses of a certain type are pre-ordered then for some customers the lens will have to be ordered through a third party which will be an added cost to the chain. Taking into account these prediction errors we use historical data to determine what should be ordered. We compare the proposed approach with the approach that was used by the chain and demonstrate significant cost savings.

The rapid growth of renewable energy sources, particularly photovoltaic (PV) panels, has led to the development of closed-loop supply chains that aim to manage end-of-life materials, such as used panels, in a sustainable manner. However, effective supply chain optimization for PV panels faces significant challenges due to the unpredictability of demand and fluctuations in commodity prices. This paper addresses these challenges by proposing a deep reinforcement learning (DRL) based predictive model to enhance forecasting accuracy in closed-loop supply chains for PV panels. The objective of this research is to develop an intelligent model capable of forecasting demand and commodity prices to improve supply chain decision-making processes. The methodology involves feature extraction using advanced convolutional neural networks (ResNet50 and DefenseNet) to capture relevant patterns from historical data, followed by DRL techniques to model the forecasting process. The model is validated using real-world demand and price data obtained from open-access sources, reported on a monthly basis. The findings indicate that the DRL-based model significantly improves forecasting accuracy compared to traditional models, offering reliable approach to optimize supply chain operations.

In light of increasing regulatory requirements and international climate ambitions, the decarbonization of global supply chains is gaining strategic importance. The manufacturing sector is central to this transformation, given its substantial contribution to global greenhouse gas emissions. This study aims to develop a practical model for evaluating suppliers according to their emissions performance. A structured set of evaluation criteria was derived from a systematic literature review, analysis of established sustainability standards, and five expert interviews. These criteria were consolidated into a structured model capturing key indicators of a company's decarbonization efforts and subsequently translated into a supplier evaluation tool. By enabling a more transparent assessment of supplier decarbonization efforts, the model supports informed decision-making and facilitates the integration of decarbonization objectives into procurement practice. The tool enables companies to systematically collect relevant information, enhances transparency and comparability along the supply chain, and supports the integration of emissions-related criteria into procurement strategies. Validation through practitioner interviews confirmed the model's practical applicability while also highlighting key challenges, such as inconsistent PCF data and limited data availability. Overall, this work contributes to the operationalization of corporate decarbonization strategies in industrial procurement and addresses a gap between theoretical frameworks and their implementation in emissions-related supplier evaluation.

The current research has focused on the analysis of the role of reverse logistics in supply chain management operations. Reverse logistics can be regarded as an essential process of supply chain management that is associated with the movement of goods from the customers to the sellers or the manufacturers. It is associated with the return and recycling of the products. It is an essential part of supply chain operations to maintain an efficient flow of goods that is also associated with the cost-saving, value creation, and risk mitigation of the companies associated with the supply chain operations. The study has focused on the analysis of the factors influencing the implementation of reverse logistics within a company, the current trends and technologies used in the reverse logistics process, the profitability aspect of this process to the company, and finally, finding the ways to improve the process.

Rapid growth in waste electrical and electronic equipment (WEEE) presents critical environmental, economic, and social challenges globally. Due to the increase in consumption of electronics the urgency for sustainable waste management too increases. The circular economy (CE) offers a regenerative model aimed at reducing waste by keeping products and materials in use through repair, reuse, repurposing, and recycling. To extend a product's life, the simplest approach would be to repair it. Reuse and refurbishment can yield significant environmental benefits, but consumer scepticism and a lack of supportive legislation limit these initiatives. Repurposing is creative, but it has safety and legal challenges. Recycling is widespread, although it often produces lower-value materials and consumes a lot of energy. The assessed frameworks offer practical methods for integrating CE ideas into business strategies. Product development and post-consumer responsibility are impacted by EPR and eco-design, while LCA helps to measure environmental trade-offs. CLSCs offer operational models for product recovery, and behavioural change models are essential for changing user attitudes and behaviours. The review indicates that comprehensive benchmarking tools that align industry actions with CE goals are required. It also highlights how important it is to combine behavioural, policy, and technical strategies in order to achieve long-term success. By addressing these gaps, businesses may transition from linear to circular systems and transform WEEE from a waste problem into a resource opportunity.

This study presents a comparative analysis of Industry 4.0 (I4.0) technology adoption at meso and macro levels within the context of the Circular Economy (CE) and Sustainable Performance (SP). Technologies such as Artificial Intelligence (AI), the Internet of Things (IoT), Cloud Computing, and Big Data are recognized as critical enablers of CE implementation; however, limited empirical research examines their differential adoption across operational (meso) and institutional (macro) domains. To address this gap, a quantitative survey was conducted with 98 meso-level organizations and 80 macro-level institutions engaged in CE initiatives. Results show higher adoption rates of IoT, Cloud Computing, Big Data, AI, and Cybersecurity at the meso level, indicating a more advanced stage of digital transformation. Conversely, Robotics shows significantly lower adoption at the macro level, highlighting a digital divide in automation capabilities. Simulation, Additive Manufacturing, and Augmented Reality display relatively comparable usage across both levels, though overall penetration remains limited. These findings offer valuable insights into digital readiness across organizational layers, highlighting the need for tailored, level specific digital strategies. Such transformation efforts are essential to accelerate CE integration and build resilient, sustainable industrial ecosystems through the effective deployment of I4.0 technologies.

Transitioning from a linear to a circular economy (CE) in the manufacturing industry (MI) is pivotal to mitigate climate change, resource shortages and environmental degradation. Circular strategies, which highlight resource and waste minimization while focusing on product lifecycle extension, demand collaboration among organizations to ensure effective implementation. This paper explores the theoretical foundations of collaboration in CE within the MI, focusing on frameworks like System Thinking, Dynamic Capabilities, Open Innovation, and Technology Transfer. A mixed-methods approach was used to refine theoretical perspectives and develop a framework. Findings highlight the need to integrate circularity into strategies for competitiveness, innovation, and sustainability.

Fair transition towards Net Zero requires effective policy design and implementation, yet its complexity makes systematic evaluation challenging. This research introduces a data-driven framework for the systematic analysis of large corpora of energy policy documents, designed to evaluate how policy design addresses household energy vulnerability in the context of the Net Zero transition. The framework employs a size-balanced document clustering method that integrates document length considerations with keyword-based thematic classification to construct coherent policy packages. These packages are then evaluated through a vulnerability coverage assessment, allowing the quantification of policy attention to energy vulnerability. By incorporating lagged correlation analysis with household energy expenditure, the framework provides an evidence-based means of assessing whether policy design effectively supports a fair energy transition. Empirical analysis of the UK energy policy documents results demonstrates how disruptive analytical techniques can enhance policy evaluation and inform strategies for more sustainable and equitable energy systems

The systematic review of circular economy and business models highlighted the importance of practices in enhancing corporate sustainability. From eco efficient product design to comprehensive waste management, companies were able to reduce their environmental print and strengthen operational efficiency. Various circular business models to different industries and markets were identified, each contributing to resource minimization and the promotion of socially responsible practices. Despite progress, organizations still faced technological and regulatory constraints when attempting to implement circular practices. The review highlighted that continued innovation, together with targeted educational initiatives, was essential for strengthening adoption both at the organizational level and within broader society. In addition, the analysis drew attention to the relevance of public policies that actively supported the transition, particularly through incentives, regulatory frameworks, and programs designed to accelerate the move toward more sustainable and resilient economic models.

Accurate electricity demand forecasting is essential for achieving sustainable development, particularly in emerging economies where energy systems face challenges of rapid urbanization, climate variability, and data scarcity. Traditional linear and statistical models often fail to capture the complex, non-linear patterns inherent in these dynamic environments. This study applies Artificial Intelligence (AI) and Big Data analytics to predict electricity consumption in Morocco. Five forecasting techniques namely Multiple Linear Regression, Support Vector Machine, XGBoost, ARIMA, and SARIMA are evaluated using a high-resolution dataset from the city of Tetouan. Through rigorous preprocessing and performance evaluation (RMSE, MAE, R²), XGBoost emerges as the most accurate and context-sensitive model. The results highlight the importance of integrating socio-climatic variables and domain-specific features to improve forecasting performance in emerging contexts. This work offers a scalable, data-driven framework for informed energy planning, providing valuable insights for policymakers and infrastructure developers in emerging countries.

This study examines the drivers and trajectory of fertility decline in a Small Island Developing State (SID) through a generational lens, highlighting the country's transition from high to sub-replacement fertility levels. Using a Data Science based approach, we investigate how factors including education, income, religion, ethnicity, marital status, and birth year have influenced reproductive outcomes across three generational cohorts. National fertility data reveal a steady decline to a stabilized rate below the replacement threshold of 2.1 around the late 1990s. Regression analysis shows that while fertility in the traditional generation was uniformly high, the transitional cohort displayed emerging variation. In the modern generation, fertility patterns became uniformly low with higher education significantly reducing fertility. These findings suggest a shift from culturally driven fertility norms to more individualized reproductive behaviour. The study contributes to understanding fertility transitions in postcolonial contexts and offers insight into the demographic implications.

Print media continues to be more trusted by the public than online sources. Unfortunately many people have little time to read such media and would prefer audio versions while driving or exercising and generally while busy with other activities. This also holds true for the visually impaired. We introduce an automated, personalized solution for emerging markets. This system follows the current shift to digital media which allows for more personalized solutions. Many can benefit from these implementations as they seek alternative revenue sources due to the decline in the revenue sources from print media. This system leverages freely available locally hosted Large Language Models (LLMS) and Text-To-Speech (TTS) synthesis tools to provide an accessible and sustainable alternative to traditional media. We compare two methods, one using Google's cloud-based NotebookLM for summarization and personalization and another entirely locally hosted solution using the LLaMA model and a quantized version of the Orpheus Text-To-Speech model. Our implementation allows for both GPU-Based solutions as well as CPU-only solutions thus reducing the need for external cloud- based systems. We show that the local pipeline also provides comparable performance to external proprietary systems. It produces a personalized experience for users while maintaining a similar level of clarity and relevance to the proprietary system. A dual-layer evaluation approach was conducted to evaluate the technical performance of the model while also evaluating the potential to monetize an Artificial Intelligence (AI) news podcasting solution. The results showed a demand for an ad-free, subscription-based model, allowing users to have more variety in how they consume news. The proposed solution underscores the transformative potential of open AI tools and technology when used in the media space.

Coastal regions of Small Island Developing States (SIDS) face mounting risks from sea-level rise, storm surge, wave action and socio-economic pressure. Traditional vulnerability indices typically employ fixed, expert-defined weights for hazard and exposure layers, which may not adequately reflect local conditions, especially in SIDS where data scarcity and environmental variability are prevalent. This can lead to a misrepresentation of effectiveness of solutions for reducing vulnerability, particularly nature based ones that play a crucial role in these contexts. We propose a data-driven framework that leverages feature importance values derived from an XGBoost model to assign objective weights to geological and ecological features. The model was trained to predict significant wave height, and the resulting feature importance values illustrate a data-driven weighting approach that can enhance the design and refinement of coastal vulnerability indices. In addition, Shapley Additive Explanations (SHAP) values were also computed to gain deeper insights into the marginal contribution of each predictor and to validate the robustness of the feature importance results. This framework provides a way to capture the relative influence of key factors, improve the representation of nature-based solutions and support more informed adaptation planning in many countries especially SIDS.

The recently introduced Kappa-Eta algorithm has demonstrated exceptional performance in regression, classification, and other machine learning tasks. In this algorithm, each feature is associated with its own hyperparameter, which has traditionally been optimized in isolation. We show that jointly optimizing all parameters over the feature space leads to better performance due to the existence of multiple local optima. Furthermore, while independent tuning is more susceptible to the influence of outliers, joint optimization reduces these effects. We also introduce regularization and compare three approaches: independent optimization, joint optimization, and joint optimization with regularization. This combined method not only mitigates overfitting and enhances robustness against noise but also reduces computation by significantly narrowing the search space.

Motion disorders affect a significant portion of the global population, often go undiagnosed or are detected too late, primarily due to limitations in current diagnostic tools. While medications can help manage symptoms, they typically affect all muscles uniformly, leading to potential side effects such as involuntary movements and memory issues. Moreover, existing gait analysis methods lack critical software engineering features, such as ease of use, cloud integration, and real-time processing capabilities, hindering their application in clinical practice. To address this problem, this work present AI-driven cloud-IoT framework to ensure reliable data handling, secure storage, and accurate gait analysis for muscle disorder detection. The proposed framework integrates IoT devices, such as smartphones, to capture motion data, eliminating the need for specialized wearable sensors. Blockchain is used to ensure data security and integrity, allowing for secure storage and sharing of sensitive medical data in the cloud. The framework utilizes a dynamic fuzzy deep neural network (D-FDNN) to analyze the motion of lower body joints, distinguishing between normal and pathological gait patterns. To validate the proposed framework, motion-based datasets, such as those from Kaggle, are used. The findings show that the AI-driven framework effectively detects muscle disorders with high accuracy and can be accessed across various devices, making it both scalable and infrastructure-independent.

Being able to accurately infer personality traits from unstructured social media text has widespread implications, from psychological profiling to personalized digital services. Such a new study describes a deep-learning-based framework novel for automatic personality type prediction through the Myers-Briggs Type Indicator (MBTI) model. By taking advantage of BERT, a bidirectional encoder representation from Transformers constructs, the proposed architecture builds four binary classifiers as per the dimensions of MBTI - Introversion/Extraversion, Sensing/Intuition, Thinking/Feeling, and Judging/Perceiving. Different from what has already been published that is single-source data or in traditional machine learning based, this study integrates and preprocesses two heterogeneous social media datasets- PersonalityCafe and Reddit-to make them robust and more generalized. Using advanced Natural Language Processing (NLP) approaches such as lemmatization and tactical contextual data augmentation, these have been employed in mitigating class imbalance and extracting semantically richer features. This proves performance with extensive experiments in distinct metrics such as F1-score and AUC, especially on imbalance data, underscoring the ability of transformer-based models for linguistic personality assessment. It will also form the basis of scalable real-world systems of personality prediction and fruition for cross-platform psychological modeling with deep language representations.

The COVID-19 pandemic and other global disruptions have emphasized the immediate necessity for enhanced resilience within supply chains, with digitalization emerging as a crucial enabler. Among the industries leading this shift, the manufacturing sector has increasingly adopted digital technologies to enhance operational visibility, responsiveness, and continuity. However, despite this advancement, there remains limited empirical clarity on how digital tools influence disruption handling, particularly under conditions of environmental uncertainty. This study aims to fill this gap by presenting and empirically validating a conceptual model that explores the disruptions caused by supply chain digitalization (DSC) and their effects on the fundamental resilience dimensions of absorption, response, and recovery. The model also considers the moderating role of environmental uncertainty (EU). 50 supply chain experts, primarily from the manufacturing sector, were selected by stratified random sampling and given a structured questionnaire to complete. Findings reveal that DSC significantly improves disruption handling performance (DHP), with all three resilience capabilities acting as partial mediators. Environmental uncertainty further moderates the relationship, highlighting the importance of contextual factors. This study provides significant insights by showing how supply chain performance and resilience in unstable situations may be improved by strategically leveraging digital transformation, particularly in the manufacturing sector.

An important technological advancement impacting supply chain operations is the emergence of financial technologies (FinTech), including Blockchain (BC), Machine Learning (ML) and Artificial Intelligence (AI). Despite their importance, there are significant gaps in the understanding of their interactions in Supply Chain Management (SCM) domain. In practice, these technologies are fundamentally changing supply chain functions by generating vast amounts of data from diverse sources such as planning systems, online platforms and social media. Through our present study of scientific literature published between 2019 and 2024, we provide significant insights that can help both researchers and industry professionals explore how these technologies can be integrated into Supply Chain. Our findings support organizations and decision makers in optimizing SCM operations based on AI to maintain competitive advantage.

Emerging technologies in the era of digital supply chain have fostered the development of innovative strategies to enhance the last mile delivery system within logistics to improve overall efficiency within the supply chain. Among these, drone technology is gaining prominence due to its distinctive attributes such as rapid delivery capabilities, cost effectiveness and increased mobility thus providing competitive edge to the businesses to respond to customer delivery requirements. While drone technology contributes to achieving a responsive supply chain, several challenges impact their real-world application and scalability. This paper identifies the critical challenges that impact the implementation and utilisation of the drone technology within last mile delivery encompassing various external and internal factors. The findings underscore the necessity for a nuanced understanding of these multifaceted barriers to ensure successful integration of drone into the last mile delivery system for an efficient and responsive supply chain

The construction sector constantly engages in significant uncertainty and risk due to the complex nature of projects, involvement of numerous stakeholders, lengthy duration, and susceptibility to external factors such as political, economic, social, and environmental circumstances. Currently, there is a lack of comprehensive empirical evidence directly comparing the effects of these collaborative and risk-sharing mechanisms on uncertainty and risk appetite across these distinct governance models within the construction industry. This research aims to address this gap by investigating the influence of contract mechanisms emphasising collaboration and risk sharing on project uncertainty and risk appetite, and whether this influence is contingent upon the specific governance model under which a construction project is delivered.

Learning has long been recognized as fundamental to progress. Embedding supply chain sustainability learning facilitates knowledge translation to sustainable practices. The study provides an analysis of the literature at the intersection of supply chain, learning and sustainability to identify research landscape overtime and emerging research trends. A bibliometric analysis was employed for this purpose. A total of 230 articles from Web of Science database were analyzed using software package Bibliometrix® on R Studio. Descriptive analysis revealed a notable increase in research interest over the past five years. Theme-based clusters were formed namely: 1) green supply chain management, 2) strategic knowledge seekers, 3) resource nurturers; 4) exploration and exploitation learning and 5) policy and practice. Areas for further potential research were identified. Overall, the study offers an integrative review and identifies avenues for future research.