Authors - Wonseong Kim Abstract - Geopolitical risk now tends to surface in information environments well before it shows up as physical disruption or moves in market prices, and often before any policy response. Current geopolitical risk indices track the salience of news, and a separate body of work on disinformation detection looks for signals of manipulation. What neither line of work does is connect manipulated discourse to the channels through which food, energy, supply chain, sanctions, and macro-financial risks are actually transmitted. To address this gap, the paper develops a DISARM-informed large language model framework for narrative-based monitoring of disinformation-driven geopolitical risks. The framework is organised as a four-layer architecture that brings together observable manipulation signals, the classification of narrative function, mapping onto risk domains, and the construction of indicators. From these layers it derives interpretable indicators that capture manipulated risk discourse, gaps in framing, concentration of narratives, and transmission across domains. Our central claim is a methodological one: once observable manipulation, narrative function, and risk-domain mapping are represented together, unstructured multilingual media can be turned into auditable early-warning signals. We intend the result as a decision-support instrument for sustainable security monitoring, not as a means of attribution or causal estimation.
Authors - Nassour Annour Saad, Mahamat Issa Hassan, Marayi Choroma, Mahamat Atteib Ibrahim Doutoum, Djaury Dadjia Abstract - Pediatric type 1 diabetes (T1D) remains a major public health priority, especially in low-resource settings where presenting diabetic ketoacidosis is still common. This systematic review (2020–2025), conducted under PRISMA 2020 and complemented by TRIPOD/TRIPOD+AIinspired criteria for predictive models, synthesizes AI/ML work on early screening and risk stratification in children. Islet autoantibodies and genetic risk scores improve discrimination, but the literature shows substantial AUC variability depending on sample size, calibration, and validation design (single split versus repeated or family-level validation). Ensemble models often outperform classical approaches with multimodal data. We emphasize external validation, class-imbalance handling, and reproducible pipelines. The main gap remains the absence of models simultaneously integrating autoantibodies, HLA/GRS, and C-peptide.
Authors - Imane Bari, Abdellatif Aziki, Zineb Alaoui Abstract - This study analyses the relationship between digital operational risk management and the financial performance of industrial firms in the Agadir region of Morocco and investigates the role of artificial intelligence in risk governance. Using a quantitative survey of 50 industrial firms and linear regression with principal component analysis, the results show that a structured digital risk management framework, covering identification, assessment, and mitigation of threats, is positively and significantly associated with financial performance (R = 58.7%, F = 3.518, p < 0.05). Several constraints are identified, including skill shortages, limited technological resources, and insufficient digital governance culture. The study further shows that AI-based tools, through automated anomaly detection and predictive analysis, strengthen the effectiveness of risk management frameworks. These findings support the integration of AI as a technical component of operational risk governance in industrial settings.
Authors - Libero Nigro, Franco Cicirelli Abstract - This paper builds on the Hartigan-Wong (HW) algorithm for unsupervised clustering. Although basic HW comes with an intrinsic high computational cost, it is known to be a better solution than K-Means, because it is less likely to get stuck around a sub-optimal solution of the data space. The paper, in particular, proposes a variant of HW, named Evolutionary HW (E-HW), which embodies genetic concepts and favors the achievement of more accurate clustering. E-HW depends on the use of a population of candidate solutions (centroid configurations), preliminarily created. E-HW is fed by a solution extracted from the population, which gets refined (crossed) and possibly replaced (mutated) following the basic HW operations. New generations of the population then come into existence. E-HW can be repeated a certain number of times, after that, experimental results highlight that the population favors the emergence of a solution close to the optimal one. To smooth out the computational burden, many operations of E-HW are implemented in parallel Java, so as to exploit the computing benefits of modern multi-core machines. The paper demonstrates the effectiveness of E-HW by using a collection of benchmark datasets, and the clustering results are compared with those achieved by competitor algorithms.
Authors - Ilgar G. Aliyev, Konul Gafarbayli, Firangiz Mammadrzayeva Abstract - Modern parallel gas pipeline systems require intelligent and operationally reliable emergency-management mechanisms capable of distinguishing real leakage events from normal technological transients under real-time operating conditions. Although IoT- and SCADA-based monitoring technologies are widely used in modern gas transmission infrastructures, most existing systems primarily rely on threshold-based supervision or empirical data-driven methods, which often lack physical interpretability and analytical decision-making capability. This paper proposes an intelligent IoT-driven emergency-management and analytical decision-making framework for parallel gas pipelines based on the integration of digital monitoring technologies with analytical gas-dynamic modeling. The proposed cyber-physical architecture combines wireless pressure sensors, SCADA-assisted supervisory control, synchronized shut-off valves, and analytical decision algorithms to ensure real-time identification, localization, and mitigation of emergency operating modes. Analytical criteria are developed for distinguishing emergency and technological pressure variations, estimating emergency detection time, localizing the leakage coordinate, and determining the optimal activation time of interconnecting pipeline valves. The proposed framework enables rapid isolation of damaged pipeline sections while ensuring adaptive gas redistribution through intact parallel lines. Unlike conventional monitoring-based approaches, the developed methodology transforms emergency control into an analytically justified intelligent supervision mechanism capable of minimizing gas losses, preventing cascade disturbances, and improving operational sustainability. The integration of IoT-based sensing with analytical decision-making additionally improves compatibility with Industry 4.0 and digital twin concepts for future smart gas transmission infrastructures.
Authors - Kirill Kalichkin, Tatiana Gritskevich Abstract - The study is devoted to the analysis of vibration agnostics problems as a method of preventive control in the design of wheel sets of railway locomotives. The study examines vibration agnostics as a preventative control method for designing individual mechanisms and components of railway locomotive wheel sets designed for long-term, safe operation. Currently, the main issue with the mechanical drives of wheel-motor unit assemblies and motor-anchor bearing assemblies in railway locomotive wheel sets remains increased vibration during highfrequency operation. The authors analyze the prediction of potential defects using digital twins, the goal of which is to enable engineers to accurately predict solutions when similar defect signs are detected during operation of different digital twin scenarios. This enables the development of preventative measures to prevent accidents at the early stages of wheel set defect development.
Authors - Mariia Nazarkevych, Vasyl Lytvyn, Oleg Stechkevych, Hanna Nazarkevych, Roman Cholkan, Danyil Korotych Abstract - An information technology for adaptive enterprise management using weak signals has been developed, which is based on the collected information about the environment, the assessment of factors affecting the enterprise, the calculation of the indicator of the impact on the enterprise based on integral dependence, the method of detecting weak signals and predicting the state of the enterprise, which provides high sensitivity taking into account changes in the environment and increases the efficiency of enterprise management. A method of recognizing weak signals is shown, which, by comparing the permissible value with the difference between the found and predicted values of the indicator of the impact on the smart enterprise based on integral dependence, provides early detection of threats or opportunities for the smart enterprise. It is proposed to develop a smart enterprise management system using weak signals based on an integrated approach and in accordance with the following principles: systematicity; integration of computer, communication and software components; modularity; openness; compatibility; variable equipment composition.
Authors - Marlon Kulatunga, Kaavya Raigambandarage, Senali Guruge, Themiya Alwis, Amila Nuwan Senarathne, Kavinga Yapa Abeywardena Abstract - Contemporary Kubernetes deployments suffer from two fundamental shortcomings: admission control mechanisms apply static rule sets without accounting for namespace operational context, and content inspection services governed by RFC 3507 remain disconnected from the orchestration layer. This work presents an integrated four-module security framework that jointly addresses both deficiencies. A probabilistic namespace characterisation algorithm employing seven weighted indicators achieves 96.7% accuracy in determining deployment tiers, even when metadata labels are absent or deliberately misleading. A compliance-driven policy orchestrator aligned with CIS Kubernetes Benchmark controls and PCI-DSS v4.0 requirements translates a unified constraint representation into artefacts for both OPA Gatekeeper and Kyverno, attaining 99.2% cross-engine decision parity. An environment-responsive traffic manager generates tier-specific Istio routing configurations, while a custom Kubernetes operator governs content scanning pod lifecycles through a multi-dimensional wellness metric that captures security-relevant signals invisible to conventional autoscalers. Evaluation on a five-node K3s cluster demonstrates full compliance coverage across 93 benchmark controls and 28 regulatory mandates, sub-five-second failover under all disruption scenarios, and correct detection of degraded scanning capability that CPU and memory metrics alone would overlook.
Authors - Ronewa Gilbert NTHATHENI, Tumiso THULARE Abstract - The rapid advancement of digital technologies has transformed the relationship between governments and citizens, creating new opportunities for participatory governance through e-government initiatives. This study evaluates the effectiveness of online engagement platforms in promoting democratic governance in South Africa. Using a scoping review methodology, the research examines the benefits, challenges, and contextual dynamics shaping citizen participation through digital platforms. Findings suggest that while online engagement tools enhance transparency, accountability, and access to information, their effectiveness is constrained by structural barriers such as the digital divide, limited institutional capacity, and low digital literacy. The study concludes that the success of e-participation initiatives depends on inclusive design, infrastructure investment, and meaningful government responsiveness. Recommendations are provided to strengthen digital governance and improve citizen engagement outcomes.
Authors - Supriya Narad Abstract - Agricultural economies are predominantly relevant in developing countries, wherein the farmers have to struggle operating under the impact of several constraints posed by crop diseases. Among food crops, the potato is a major one with vulnerable destructive diseases like Early Blight and Late Blight, capable of destroying the yield if detected late. Old methods of visual inspection are time-consuming and sometimes erroneous because of laxity, human error, and lack of expertise. With this research, an automated intelligent disease detection system is devised, making use of image processing and deep learning, Arduino, specifically Convolutional Neural Networks (CNNs). The model was trained using potato leaf images from the Plant Village dataset, which are improved using various preprocessing techniques, including color space conversion, image augmentation, and image resizing. The proposed CNN architecture achieved a high rate of classification accuracy of 97.2% in distinguishing healthy leaves vs. Early blight and Late blight infected leaves. Lightweight, reliable, and fast, it supports implementation on mobile or handheld devices in low-resource environments, thus giving farmers the ability to use them for timely diagnostics. The system has good prospects for scaling up for other crops and disease types in future versions.
Authors - Khadidje OUSMANE KOSSI, Mandicou BA, Bachar Haggar SALIM, Simon Antoine SARR, Maboury DIAO, Alassane BAH Abstract - Heart disease in athletes remains a significant challenge in sports cardiology and an important public health concern, particularly among young competitive individuals at risk of sudden cardiac events. Although pre-participation screening programs are widely implemented, diagnostic uncertainty persists, especially in distinguishing physiological cardiac remodeling from pathological cardiomyopathy. This complexity results from the interaction of genetic predisposition, structural adaptation, electrophysiological variability, and cumulative training exposure. Using the PRISMA framework, this study presents a systematic review of research published between 2015 and 2025 to evaluate the application of artificial intelligence (AI) in the diagnosis and monitoring of cardiovascular diseases in athletes. The analysis reveals that most studies rely on unimodal, monocentric, and retrospective designs, often based on limited datasets and lacking external validation. Despite high reported performance metrics, performance degradation of 5–10% in external cohorts is frequently observed. Furthermore, explainability techniques are inconsistently applied, and real-world clinical integration remains limited. Only a small number of studies adopt multimodal approaches integrating electrophysiological, imaging, biological, and training-related data. These limitations restrict the clinical translation of AI models. Future research should prioritize multicenter, diverse, and explainable multimodal frameworks to support reliable cardiovascular risk stratification and return-to-play decision making.
Authors - Oluwaranti A. Omowami Abstract - Work-related musculoskeletal disorders (WMSDs) are among the most prevalent occupational injuries in construction, driven by heavy lifting, awkward postures, repetitive motion, and whole-body vibration. Traditional ergonomic risk assessment methods are retrospective and unable to capture the dynamic conditions of construction sites. Wearable sensor technologies offer a real-time, objective alternative. This structured narrative review examines the implementation, accuracy, and occupational health outcomes of wearable sensor systems applied to ergonomic risk monitoring among construction workers. A structured review of peer-reviewed literature from 2017 to 2024 identified six sensor categories: inertial measurement units (IMUs), wearable insole pressure systems, surface electromyography (sEMG), electrodermal activity (EDA) sensors, heart rate monitors, and smartphone embedded sensors. Reported posture classification accuracy reached up to 99.01% under controlled conditions using deep learning classifiers. Key implementation barriers include sensor discomfort, motion artifacts, worker acceptance, data privacy and cybersecurity concerns, and the multi-employer structure of construction. A consistent gap exists between laboratory validation accuracy and real-world field performance. Occupational health outcome studies remain limited. Future priorities include longitudinal field validation and integration with behavior-based safety frameworks.
Authors - Lucian Julio Felix da Costa, Claudio de Souza Baptista, Andre Luiz Firmino Alves Abstract - Auditing public procurement processes is essential to ensure transparency, accountability, and efficiency in the management of public funds. However, the increasing complexity of procurement procedures poses significant challenges for auditors, particularly regarding the timely detection of pricing irregularities. This paper presents a software tool for price comparison designed to support the identification of overpricing and underpricing in public works procurement. The proposed solution leverages semantic retrieval and historical price comparison techniques to analyze procurement data and integrate up-to-date market information. Additionally, the application provides interactive visualizations and semantic retrieval mechanisms to support auditors during procurement price analysis activities. The expected contribution of this study lies in improving the effectiveness and accuracy of procurement oversight, strengthening financial analysis processes, and contributing to the prevention and deterrence of fraudulent practices among bidders.
Authors - Vanderson dos Santos Araujo, Eliane Tamara Lima Oliveira, Pedro Manoel Hermínio Alves, Andre Luiz Firmino Alves, Claudio de Souza Baptista Abstract - Auditing public tenders requires analyzing lengthy documents to verify compliance with tender notices, a time-consuming task prone to human error. This article empirically evaluates the use of Large Language Models (LLMs) to assist with auditing tender notices. A total of 50 official tender notices and 736 audit instances were analyzed, comparing three context-provisioning strategies: expanded context windows, integrated file retrieval, and a custom Retrieval-Augmented Generation (RAG) pipeline. The results show that no single approach is superior across all scenarios. Models with long windows performed better at confirming explicit conformities, whereas retrieval-based strategies demonstrated greater sensitivity to potential non-conformities due to omissions. The analysis also indicates that the type of question strongly influences performance, especially for interpretive questions or those that rely on the absence of documentary evidence. As a key contribution, the study demonstrates that the effectiveness of AI-assisted auditing depends on the combination of the contextualization strategy, the quality of the retrieved context, and the formulation of the questions, reinforcing the role of LLMs as tools to support the auditor.
Authors - Ahmed Belgaid Abstract - Digital transformation has profoundly altered the organization of work and reinforced the importance of mastering digital tools for employability and productivity. In this context, this study highlights the challenges of digital accessibility for employees with visual impairments and its application through an analysis of the new WCAG standards. The aim of our analysis is to demonstrate that digital accessibility consists of guaranteeing an inclusive digital transformation; it is not limited to simply ac-quiring digital solutions or adapting existing ones. It is a comprehensive preparation process and an integrated approach involving various stakeholders within the company.
Authors - BENABDELLAH Nouhaila, CHRAIBI Abdeslam, BENRREZZOUQ Rhizlane Abstract - The importance of smart logistics dashboards as key tools for digital transformation in manufacturing companies is becoming more recognized. They enable real-time visibility, in-tegrate various data sources, and allow for analysis. However, there is limited research on how these dashboards affect decision quality and operational performance in emerging countries. This paper explores the impact of smart logistics dashboards on data-driven decision making (DDDM) and operational performance in Moroccan manufacturing firms. We conducted a quantitative survey among logistics and operations managers and examined a proposed concep-tual framework using PLS-SEM. The findings showed that the capabilities provided by smart dashboards significantly improve decision quality through better data integration and real-time analytics. Additionally, DDDM plays a key role in the relationship between smart dashboards and operational performance. These results are important for the field of smart logistics and digital transformation and have valuable practical implications for management in manufactur-ing firms in emerging countries.
Authors - Francisco Castro Murillo, Teresita de Jesus Alvarez Robles, Andres Sandoval Bringas, Monica Carreno Leon, Francisco Javier Alvarez Rodriguez Abstract - Web accessibility for visually impaired users remains a critical challenge in HCI. While technical compliance with WCAG is well-established, knowledge of User Experience (UX) evaluation methods tailored to non-visual interaction is fragmented. This paper presents a systematic mapping of the literature (2016–2026), analyzing 18 high-quality studies selected from 134 records retrieved from IEEE Xplore, ACM DL, SpringerLink, and ScienceDirect. Results show that user testing is the predominant method, often combined with standardized questionnaires like SUS and NASA-TLX. However, critical gaps persist: small sample sizes, inconsistent participant reporting, and a lack of metrics designed for non-visual interaction. This study contributes a taxonomy of visually impaired user profiles and identifies the technical requirements for building an inclusive web usability laboratory. By bridging the gap between technical auditing and real-world user satisfaction, this work provides a roadmap for the development of the MEUX LAB, ensuring more equitable digital evaluation environments.
Authors - Vikas Pandey, Sanasam Chanu Inunganbi Abstract - Plagiarism has become a serious problem in universities, research organizations, and professional workplaces, affecting academic integrity and the originality of work. While cloud-based detection tools are widely used, they share a fundamental problem that rarely gets discussed openly: every document is required to be submitted and handed over to a third-party server. For unpublished research, legal drafts, or any sensitive material, this trade-off is not acceptable. The proposed method, named ZERO, takes a different approach and runs entirely on the local machine, watching the clipboard quietly in the background and scoring text against a local TF-IDF corpus in under 200 milliseconds, with no uploads, no accounts, and no data leaving the device. An optional web scanning module is available when broader source coverage is needed. On top of the similarity score, ZERO provides a word-level risk heatmap, a sentence-by-sentence originality breakdown, a stylometric module called Writing DNA, and a scan history timeline. Testing on 60 hand-labelled samples showed that a recalibrated scoring curve brings the average score on original technical writing down from 34.7% to 9.8%, while keeping verbatim-copy detection at 95%. API credentials are stored in the OS keychain, and inter-process communication is locked to a strict channel whitelist, making the application well-suited for confidential and pre-publication work.
Authors - Sanae.Chakir, Adil.Bekraoui, El moukhtar Zemmouri, Hassan.Majdoubi, Mhamed. Mouqallid Abstract - For cuttings to successfully root indoor environmental conditions in a greenhouse are essential. This article examines the efficacy of two predictive models, linear regression and artificial neural networks, in predicting the parameters associated with rooting plant cuttings. For evaluation the analysis uses the RMSE MAPE and R² indices. According to the results artificial neural networks perform better than linear regression in terms of prediction accuracy. By utilizing these insights, farmers can use artificial neural network models to implement optimal control strategies which will al-low them to accurately predict indoor variables and ultimately increase crop productivity.
Authors - Miroslav Stefanov, Stoyan Denchev, Kristiyan Stefanov Abstract - Large Language Models (LLMs) are increasingly used in educational settings, but they are not inherently constrained to the boundaries of specific instructional materials. This can lead to unsupported claims, external knowledge leakage, and reduced instructional precision. This study proposes and evaluates a lesson-bounded LLM architecture for reliable educational AI systems. The architecture combines retrieval-augmented generation, context restriction, structured response control, explicit refusal behavior, and post-hoc confidence calibration. Using a multi-domain instructional dataset and a benchmark of inscope and out-of-scope questions, the proposed system is compared against an unconstrained baseline LLM. Results show strong retrieval discrimination and boundary control, with high Area Under the Receiver Operating Characteristic Curve, high Average Precision, strong refusal recall, low out-of-scope answer rate, reduced verbosity, and improved support-based instructional density. Calibration analysis further shows that raw retrieval scores are not reliable probability estimates, but Platt scaling substantially improves confidence reliability. These findings suggest that lesson-bounded architectural constraints can improve the controllability, auditability, and reliability of intelligent educational systems while highlighting the need for stronger factuality evaluation and confidence interpretation.
Authors - Hatem Yousif Alkhonini, Fethi Fkih Abstract - This study evaluates five post hoc explanation methods using XAI-Bench under controlled settings with ground-truth explanations. Results show significant differences in robustness, with MAPLE outperforming Shapley-based methods under high correlation. Feature correlation impacts explanation quality more than the choice of method, and performance degrades with increasing dimensionality-especially for LIME. Exact methods become infeasible beyond d=10, and robust evaluation requires multi-seed replication.
Authors - Maria Jose Cantos Cedeno, Kevin Michael Mero Ramirez Abstract - The spread of disinformation through social media, messaging apps, and other digital channels is a growing problem in Ecuador, due to the limitations of manual fact-checking processes in the face of the high volume of information. In this context, Transformer-based models are presented as high-potential solutions for detecting fake news across various domains and languages. The objective is to comparatively evaluate Transformer architectures pre-trained using fine-tuning techniques for the automatic classification of fake and real news in the Ecuadorian context. The CRISP-DM methodological framework was applied to guide the development of deep learning models. A balanced dataset of 5,000 news items in Ecuadorian Spanish was constructed, equally distributed between real and fake news. Data processing was carried out through a 12-stage sequential pipeline to reduce noise, prevent data leakage, and preserve relevant semantic features of Spanish. Furthermore, the models were trained using homogeneous hyperparameters and evaluated using various metrics employed in the scientific field. As a result, all models exceeded 89% accuracy; BETO achieved the highest precision, mBERT the best recall, and DistilBERT the highest computational efficiency. It is concluded that Transformer architectures proved to be scalable, effective, and viable solutions for the automatic detection of fake news in Ecuador.
Authors - Arsalan Vahi Abstract - The successful deployment of the Internet of Things (IoT) applications relies heavily on their robust security, and lightweight cryptography is considered an emerging solution in this context. While existing surveys have been examining lightweight cryptographic techniques from the perspective of hardware and software implementations or performance evaluation, there is a significant gap in addressing different security aspects, such as design principles, specific to the IoT environment. This study aims to bridge this gap. This research presents an examination with focusing on the security evaluation of symmetric lightweight ciphers commonly used in IoT systems. The objective of this study is to provide a concise overview of lightweight ciphers with emphasizing on their security challenges which is an essential consideration for real-time and resource-constrained applications.
Authors - Amine El Ameri, Ahmed Drissi Abstract - Quantum computing threatens TLS through the Harvest Now, Decrypt Later (HNDL) attack: adversaries record encrypted traffic today and decrypt it once quantum capabilities mature. Existing approaches integrate post-quantum key encapsulation mechanisms directly into the TLS handshake; while cryptographically sound, they inflate the first handshake message and cause IP fragmentation of the ClientHello record that some middleboxes reject, leaving these solutions undeployable on today’s Internet. This paper proposes PH-PQ-TLS, a post-quantum key establishment framework for TLS 1.3 that adds post-quantum confidentiality while preserving the standard handshake, thereby avoiding fragmentation and middlebox incompatibilities. The framework requires no redesign of TLS, uses only standardized TLS mechanics, and is crypto-agile by design. We provide a full Go implementation and evaluate its overhead against a standard TLS 1.3 baseline. The ClientHello record measures 289 bytes, well below the minimum IPv6 Maximum Transmission Unit of 1280 bytes, whereas a hybrid ML-KEM-768 ClientHello reaches roughly 1473 bytes and exceeds that threshold. The post-handshake upgrade adds 1.18 ms of latency per full connection, a one-time cost amortized over the session lifetime and avoided on Pre-Shared Key resumed connections. These results show that HNDL protection does not require redesigning TLS, replacing the Web PKI, or accepting deployment failures, offering a practical, incremental path toward quantum-resilient TLS on today’s infrastructure.
Authors - Sayuni Dewapriya, Pehan Gunasekara, Shenal Peiris, Charith Herath, Kavinga Yapa Abeywardena, Ayesha Wijesooriya Abstract - DNS is often trusted within modern network environments, making it a common channel for covert communication, malware activity, and infrastructure abuse. This paper presents a near-real-time AI/ML-based DNS threat detection framework designed to identify suspicious DNS behaviour and transform raw network activity into actionable security events. The proposed approach combines machine learning, behavioural analysis, flow-based detection, event aggregation, risk scoring, and contextual threat intelligence to improve visibility across plaintext DNS and DNS-over-HTTPS traffic patterns. The framework supports practical security operations by reducing raw alert noise and producing structured outputs suitable for dashboard monitoring and SIEM-based investigation. Evaluation using public datasets, generated attack traffic, and live DNS traffic demonstrates that the framework can support effective DNS threat monitoring, alert prioritisation, and SOC-level analysis.
