The European Journal of Research and Development

Improving In-Vehicle Air Quality with Bio-Additive ABS Composites

Songül Kılınç, Ümmet Ayyıldız — Fri, 12 Dec 2025 00:00:00 +0300

In recent years, significant research efforts have focused on improving indoor air quality in vehicles and reducing volatile organic compound (VOC) emissions. Interior trim components release harmful gases, particularly under elevated temperature conditions, due to the degradation of organic structures, posing health risks to passengers. This risk is especially critical for children and animals who are exposed to prolonged travel periods in service vehicles. In this study, bio-based additives were incorporated into recycled acrylonitrile butadiene styrene (ABS) matrices used in interior trim sheet production to reduce environmental impacts and improve thermal performance. A mixture obtained from marine-origin algae and terrestrial plant powders (nettle, oak, and poplar leaves) was added to recycled ABS at 2 wt%. Total Organic Carbon (TOC) measurements were conducted under ambient conditions. Results showed that carbon emissions from bio-additive plates were 87.7% lower than those from non-additive plates. These findings demonstrate that natural additives exhibit gas adsorption capabilities within ABS matrices and offer an effective and sustainable alternative for improving in-vehicle air quality. In this context, the present work provides an important contribution both to recycling-based polymer utilization and to the development of eco-friendly, bio-composite automotive materials.

Analytical Prediction and Experimental Validation of Bolt Self-Loosening under Vibration

Can İçmez, Umut İnce, Samed Enser — Wed, 03 Dec 2025 00:00:00 +0300

The self-loosening of bolted joints under vibrational loading remains a persistent challenge in many engineering applications, especially in the automotive industry, where safety and reliability are of paramount importance. Predicting self-loosening behavior is challenging because numerous parameters influence joint performance, as well as the limitations of conventional experimental testing. This study presents a novel analytical model for predicting bolt and nut loosening behavior under transverse vibration. The model extends existing approaches by incorporating additional parameters such as displacement, clamping force, and under-head friction torque. To enhance usability, the model was implemented in an MS Excel–based calculator with macro functions, enabling engineers to perform loosening analyses under varying conditions. The model adapts and extends existing approaches from the literature by incorporating an energy equilibrium approach, which calculates bolt rotation by balancing the torsional strain energy accumulated during vibration with the kinetic energy released once the applied torque exceeds the critical threshold. The analytical predictions were validated through Junker vibration tests, showing strong agreement with experimental data. The proposed model and tool provide a practical and accessible method for predicting loosening, thereby enabling the design of safer and more reliable fasteners while strengthening industrial competitiveness.

A New Approach Based on Ensemble Clustering for the Fabric Color Batching Problem

Yusuf Kuvvetli, Ebru Çalışkan, Onur Balcı, Esra Tabaş Asiltürk — Fri, 05 Dec 2025 00:00:00 +0300

The fashion industry is one of the industries most influenced by aesthetics and quality. This necessitates that products manufactured for this industry possess high quality and aesthetic appeal. Denim products are among the most frequently used in this industry for various purposes. This study proposes an ensemble clustering approach for visually sorting batches to reliably classify color consistency in denim fabrics. First, separate batches were obtained using three common methods (DBSCAN, hierarchical clustering, and K-Means) with 800×800 pixel RGB images of fabric samples for each order. Then, an ensemble rule based on the majority principle was designed to reduce inconsistencies between methods and balance random initialization and parameter sensitivity. Each sample was assigned to the final batch according to the majority preference among the batches given by the three algorithms. It is evaluated that the proposed approach by comparing it with reference batch assignments predefined by experts. The outputs of the individual algorithms and the ensemble results are compared each other. The findings show that the ensemble rule produces more stable batches that are closer to expert decisions. While preserving the strengths of the individual methods, the ensemble rule reduces the impact of their weaknesses.

Synthesis and Characterization of Stereoselective Ozonides for Sustainable Textile Wet Processes

Gökhan Ceyhan, Orhan Işık, Onur Balcı — Sat, 06 Dec 2025 00:00:00 +0300

This study focuses on the stereoselective synthesis of ozonides for potential applications in different textiles processes. By synthesizing controlled ozonide in a closed-circuit reactor, high stereoselectivity (>90% ozonide formation) was achieved, enabling sustainable denim fading, cotton bleaching and/or different textile washing processes. This method will significantly reduce water usage and chemical discharge compared to traditional processes. In-depth analyses using UV-vis spectroscopy, FTIR, NMR, X-ray diffraction (XRD), cyclic voltammetry (CV), and oxidation-reduction potential (OPR) measurements demonstrate selective chromophore degradation without cellulose degradation, confirming the role of ozonide intermediates in targeted oxidation. This innovation aligns with the EU Green Deal principles, which promote circular economy applications in textiles. Its scalability and low energy profile highlight its applicability for eco-efficient textile production.

Development of a New Door System with High Thermal Resistance and Improved Sealing Performance for Refrigerated Display Cabinets

Sun, 07 Dec 2025 00:00:00 +0300

In this study, a new door system design was developed for the door-to-door and door-to-frame junctions of the Refrigerated Display Cabinet (RDC), where energy savings are concentrated. The aim is to improve sealing performance by reducing thermal bridges, thereby reducing the system's overall energy consumption. As part of the design, the silicone filling volume in the door bottom mold was increased, a magnetic seal element was integrated, and a new door gasket was developed. Test studies were conducted in accordance with ISO 23953-2:2023, and comparative analyses were performed using the current system. Experimental results showed that the new design improved heat transfer efficiency by increasing the temperature difference between the evaporator inlet and outlet. Additionally, the average product temperature in the cabinet was improved by up to 5%. According to energy consumption analyses, annual energy consumption decreased from 13,231 kWh to 6,906 kWh, resulting in approximately 47.8% energy savings. Carbon emissions calculations over a ten-year lifespan showed a decrease from 82,033 kg of CO₂ to 42,816 kg of CO₂. As a result, the new door system was evaluated as a long-lasting solution that increases energy efficiency, contributes to environmental sustainability, and provides a more sustainable solution.

Machine Learning-Based Vehicle Renewal Prediction: A Hybrid Approach for Customer Retention in Premium Automotive Markets

Selçuk Bayracı, Garen Bozoğlanoğlu, Turgay Tugay Bilgin — Sun, 07 Dec 2025 00:00:00 +0300

Customer retention and vehicle renewal prediction remain critical challenges in premium automotive markets. This study presents a comprehensive data-driven framework for predicting BMW customer renewal probability using historical transactional and behavioral data from Borusan Otomotiv's enterprise systems. We developed a hybrid machine learning model that integrates Random Forest feature selection with Binary Logistic Regression to achieve interpretability while maintaining predictive accuracy. The model leverages customer demographics, service engagement metrics, and ownership patterns to generate individual-level renewal probability scores.

Evaluated on 1,211 holdout observations through temporal validation, the model achieved 77% overall accuracy and an AUC-ROC of 0.80, demonstrating strong discriminatory power in distinguishing between renewal and non-renewal customers. Model outputs are transformed into five operational risk grades (G1-G5) and seamlessly integrated into Salesforce CRM, enabling proactive customer relationship management and targeted retention strategies.

Key empirical findings indicate that service expenditure patterns, time since last purchase, and multi-vehicle ownership significantly influence renewal likelihood. The framework bridges predictive analytics with operational deployment through automated data pipelines and continuous model monitoring, representing a practical approach to data-driven customer retention in the automotive sector.

UWB-Based High-Precision Real-Time Positioning and Multi-Dimensional Visualization

Onur Yılmaz, Turgut Aydoğdu, Hasan Berkhan Özkan, Savas Barış, Yusuf Kaya — Wed, 10 Dec 2025 00:00:00 +0300

This study presents a high-precision indoor positioning and multi-dimensional visualization system, named Virtual Positioning System (VPS), which utilizes Ultra-Wideband (UWB) technology. The VPS features an integrated architecture comprising portable Tag devices, fixed anchor units, a data collector called Position Box, and a web-based server. Tests conducted in various scenarios (office, factory, and retail environments) demonstrated that the system achieves a positioning accuracy of ±30 cm and provides high data stability.

The Two-Way Ranging (TWR) algorithm and Kalman filter minimize measurement noise, while IEEE 802.3-based communication prevents data loss. The 2D and 3D visualization modules provide capabilities for movement tracking, density mapping, and area-based analysis. In particular, 3D visualization enhances operational awareness by providing depth perception in multi-story buildings or metal-dense environments. The VPS is well-suited for future developments in terms of energy efficiency, signal stability, and visualization performance, adding value to industrial, corporate, and security-focused operations.

CarrGo® Sorter: Parcel Sorting System with Autonomous Multi-Robots

Wed, 10 Dec 2025 00:00:00 +0300

This paper presents the design, development, and validation of the CarrGo^® Sorter System, an autonomous multi-robot sorting system for logistics and cargo handling environments. The system automates parcel classification, routing, and transferring operations using a fleet of Automated Guided Vehicles (AGVs) controlled by a software. Each robot is equipped with embedded sensors, magnetic line follower, and RFID-based localization to navigate a structured grid platform. By integrating barcode reader and real-time communication protocols, the CarrGo^® Sorter System achieves high sorting throughput with reduced dependency on human operators. Simulation and field tests demonstrated that the system can increase sorting speed while avoiding collision risk through centralized traffic management. The results support the potential of multi-agent robotic platforms to improve intralogistics by combining embedded control, intelligent coordination, and autonomous navigation.

Effect of Amorphous Silica–Forming Additive on Porosity and Mechanical Strength in Autoclaved Aerated Concrete Thermal Insulation Board

Yunus Ion Grecu, Ezgi Biçer, Emre Fenerci, Ebru Erdoğan, Fatma Bakır — Fri, 12 Dec 2025 00:00:00 +0300

Autoclaved aerated concrete (AAC) thermal insulation board has a density of 130–155 kg/m³, a compressive strength above 0.4 MPa and a thermal conductivity value of 0.045 W/m.K. It is a Class A non-combustible, mineral-based and non-toxic material and used for thermal insulation from the outside, inside, in the middle, underground, on floors, and roof surfaces. The porous structure of the material decisively affects its mechanical and thermal conductivity properties. In this study, the potential for pore size reduction was evaluated by adding ratios of 0%, 0.1%, 0.25%, 0.5%, 0.75% and 1% amorphous silica-forming additive to the AAC thermal insulation board by mass. Furthermore, the mechanical performance was compared with the corresponding pore size characteristics. In determining the pore distribution, the air pores in the structure were examined by image analysis technique based on the Monte Carlo approach. When the density and compressive strength of the samples obtained after hydrothermal curing were compared with the A value, it was observed that the highest increase was 29.94% with a 1% additive rate. Scanning electron microscope (SEM) and X-ray diffraction (XRD) analyses showed that the amount of tobermorite increased continuously up to a dosage of 0.5%. The fact that the addition of the admixture by mass reduces the pore diameter, reduces density and increases compressive strength reveals that the amorphous silica-forming additive is usable in AAC thermal insulation board. Achieving the same compressive strength with less material during the production phase and reducing per-unit energy consumption during service due to improved thermal insulation associated with smaller pore sizes are critical for lowering the carbon footprint.

Secure Use of Artificial Intelligence with Artificial Intelligence Based Control

Fatih Mehmed Bilgin, Ali Aydın, Tugberk Zurnacı, Engin Bilici — Fri, 12 Dec 2025 00:00:00 +0300

Artificial intelligence applications have increased in recent years, providing benefits that increase the productivity of individuals and organizations. Individuals and organizations consult with AI tools in many areas, seek their assistance, and create value using these tools. However, the use of AI tools brings with it various security concerns. Open-source AIs have higher capabilities than those hosted on-premise environments. This encourages individuals and organizations to use open-source or paid versions. This study aims to identify and prevent unauthorized sharing of potentially sensitive data with third parties during paid or open-source use of AI tools using AI-assisted detection and prevention. The study, aims to use a combination of natural language processing, big data, and machine learning methods during detection processes, will also focus on customizing the models to be organizations or person-focused, in addition to general sensitive data, and increasing success in capturing sensitive data by fine-tuning the models. It will enable the implementation of blocking or masking processes after a successful detection process.

A Compact Non-Intrusive Measurement System for Critical Dimensions and Calibration Chart Generation of Underground Fuel Tanks

İlker Değirmencioğlu, Savaş Barış, Yusuf Kaya — Wed, 31 Dec 2025 00:00:00 +0300

This paper presents a compact, non-intrusive measurement system designed for determining the critical dimensions and generating calibration charts of underground fuel tanks via a 2-inch access port. The system employs a laser electronic distance measurement (EDM) device located outside the Zone 0 hazardous environment, with the beam directed into the tank through a mirror-based tilt mechanism. A key contribution is the ability to generate accurate calibration charts. Mirror tilt actuation is controlled via a linear actuator, where the non-linear relation between displacement and angular rotation can be resolved either through a lookup table or analytically as the mechanical linkage properties are known. The methodology involves coarse scanning for tank geometry estimation followed by targeted high-resolution scans at critical angles to derive diameter, length, dome geometry, and inclination. Real-world results demonstrate volumetric accuracy better than 0.2%, with an expected performance of 0.3% [1] in calibration chart generation, confirming that the system meets industrial standards for underground storage tank metrology, including OIML R 71.

Development of Yarn Detection Sensor for Circular Patterned Yarn Dyeing Machine

Neslihan Okyay, Fatih Işık, Necip Fazıl Ateş, Sümeyye Kes — Sat, 29 Nov 2025 00:00:00 +0300

The textile industry has undergone a radical transformation in recent years, driven by digitalization, automation, and the pursuit of sustainable production. In this transformation, electronic and computer-based sensor technologies are gaining prominence in critical areas such as production line monitoring, process control, quality assurance, and energy efficiency. This study examines the development of a yarn detection sensor for a circular machine featuring patterned yarn dyeing technology. Unlike traditional dyeing methods, this technology combines the yarn transfer process with the dye spray system, enabling direct patterning of the yarn. Only the required amount of dye molecules chemically reacts with the yarn, resulting in significant savings in water and energy consumption. The system, with 36 independent dyeing stations, offers flexible production; however, yarn breaks, resulting from factors such as yarn raw material, strength, twist, and friction, lead to production losses and defective package formation. The import of currently used yarn detection sensors poses a significant disadvantage in terms of cost and lead time. Therefore, this study has developed a domestically produced sensor that can instantly detect yarn breaks, communicate with the machine in real time, and automatically stop the station. The developed system will minimize production losses and delivery delays, saving energy and resources. Consequently, the design and integration of a domestic yarn detection sensor will not only improve production quality and efficiency, but will also contribute to reducing external dependency and promoting environmental sustainability. In this respect, the study can contribute to advancements in smart production technologies in the textile industry.

Design and Development of a Customer Data Platform for Loyalty Programs: Data Deduplication and Personalized Marketing Infrastructure

Gizem Akman Köksal, Erhan Efe, Uğur Serkan Taşkın — Sun, 14 Dec 2025 00:00:00 +0300

This position paper presents the architecture and deployment of a Customer Data Platform (CDP) for the Koçtaş loyalty program to enhance data quality, unification, and personalization-based marketing strategies. The project entails bringing together disparate customer data collected across multiple channels into a single, deduplicated data store to enable advanced analytics and AI-driven personalization. By employing a combination of big data technologies, cloud infrastructure, and machine learning algorithms, the proposed system will enable real-time data processing of information, customer segmentation, and predictive modeling. Through this system, the platform will enhance marketing performance, customer satisfaction, and operational efficiency while adhering to data privacy legislations such as GDPR and KVKK compliance. The article situates the project within the contexts of customer relationship management (CRM), loyalty program studies, and personalization studies. It discusses data consolidation, deduplication, and system development processes, highlighting innovative elements such as adaptive algorithms, real-time learning processes, and secure data management. Expected gains are increased marketing ROI, additional loyal customers, and streamlined operational processes. The paper concludes with the analysis of the long-term potential contribution of the project and with future research avenues for large-scale data-driven marketing infrastructures.

A Multimodal Deep Learning Framework for Predicting Machine Anomalies Using IoT-Enabled Vibration and Sound Data

Alper Saylam, Mehmet Ayberk Cakar, Haluk Atlı — Mon, 15 Dec 2025 00:00:00 +0300

Unplanned machine downtimes caused by component failures, overheating, or mechanical stress significantly impact manufacturing efficiency and profitability. Predicting such failures before they occur is a core objective of smart manufacturing and Industry 4.0. Leveraging recent advances in sensor technology and machine learning, this study proposes an anomaly detection architecture that predicts the operational state of manufacturing machines one step ahead, enabling early detection of potential downtime.

The system integrates two primary data sources: vibration signals collected by an IoT-enabled device and sound recordings obtained from a microphone positioned close to the manufacturing equipment. These complementary signals capture the machine’s dynamic behaviour under varying operational conditions. While vibration and line status data are directly utilized, sound recordings undergo pre-processing using a low-pass filter to remove irrelevant background noise. The filtered recordings are segmented into one-minute intervals, and statistical features are extracted in both time and frequency domains, including mean, standard deviation, skewness, and kurtosis. Since the available dataset covers only one day, a moving block bootstrap technique is employed to improve robustness and generalization.

Two deep learning architecture, Long Short-Term Memory (LSTM) and Multi-Layer Perceptron (MLP), are implemented to forecast the machine state at time t + 1. The dataset, consisting of nine features and approximately 13,200 samples, is divided into training, validation, and test sets in a 70/15/15 ratio. Both models are trained using the Adam optimizer and binary cross-entropy loss. Performance is evaluated using precision, recall, and F1 score metrics.

Overall, the proposed approach demonstrates that combining vibration and acoustic data with deep learning can effectively predict machine anomalies in real time, contributing to proactive maintenance and reduced production downtime in smart manufacturing environments.

Probability-Calibrated Ensemble Methods for Automotive CRM Lead Scoring

Bilal Sedef, Selçuk Bayracı, Turgay Tugay Bilgin — Wed, 17 Dec 2025 00:00:00 +0300

Accurately predicting sales conversion in automotive CRM systems is critical for optimizing marketing spend and sales team efficiency. This study presents a calibrated ensemble framework combining XGBoost, Gradient Boosting, and Random Forest classifiers to predict lead conversion probability in automotive dealership operations. Using 62,859 real-world leads collected between July 2024 and July 2025, we developed a systematic pipeline encompassing behavioral feature engineering, statistical feature selection, ensemble modeling, and probability calibration via Platt scaling. The calibrated ensemble achieved an AUC of 0.841, Brier score of 0.146, and 19% improvement in top-decile precision over baseline logistic regression. The framework provides actionable lead segmentation into four priority tiers, directly supporting sales resource allocation and marketing campaign optimization. Results confirm that probability calibration is essential for automotive CRM applications where predicted scores inform operational decisions.

A Smart Shopping Cart: Shopper®

Wed, 17 Dec 2025 00:00:00 +0300

This study presents the design and development of Shopper^®, a smart shopping cart system that integrates embedded hardware, computer vision, and real-time localization technologies to enhance the in-store shopping experience. The system combines a custom control panel, dual-camera based barcode recognition architecture, loadcell weight tracking, and a mobile-based authentication mechanism. The Shopper^® autonomously verifies items, updates shopping cart contents, and initiates automatic checkout when reaching designated payment zones. By merging user experience design (UX) principles with embedded IoT hardware, the solution reduces queue times and enriches the consumer’s shopping journey. The results indicate a significant improvement in transaction speed and customer satisfaction, supporting the viability of smart carts as an effective bridge between physical and digital retail ecosystems.

A Temporal-Weighted Hybrid Recommender for B2B Vehicle Auctions Using Word2Vec Embeddings

Selçuk Bayracı, Uğur Barış Öztürk, Turgay Tugay Bilgin — Fri, 19 Dec 2025 00:00:00 +0300

Used car auction platforms face unique challenges in personalized recommendation due to extreme data sparsity, high inventory turnover, and real-time operational constraints. This study develops and evaluates a hybrid recommendation system combining Word2Vec embeddings for categorical vehicle attributes with standardized numerical features, applying temporal decay weighting to prioritize recent user interactions. Deployed on Azure infrastructure, the system was evaluated using 12 months of transaction data from a Turkish B2B auction platform comprising 5,322 users, 24,987 vehicles, and 1.87 million interactions. Offline evaluation demonstrates superior performance over baselines (Hit Rate@10: 0.456 vs 0.234 popularity baseline, 94.9% improvement). Production deployment over six months (April–September 2025) generated 977 recommendation-driven sales representing 15.26% of total platform transactions and 17.27M TL in commission revenue. Quasi-experimental analysis revealed a 26.7% increase in monthly purchase frequency among active users, yielding 420 incremental transactions. Results demonstrate how interpretable temporal-weighted embedding models generate measurable commercial value in high-turnover, data-sparse B2B marketplaces.

Pressure-Controlled Runner Optimization and Filling Balance Analysis in Multi-Cavity Injection Molds

Muhammet Furkan Çalık, Ceren Giray Karaeli — Sat, 20 Dec 2025 00:00:00 +0300

In this study, the effects of runner-system design on filling balance and pressure distribution in multi-cavity injection molds were investigated through Moldex3D simulations. Four runner configurations—H-type, Symmetrical-type, Star-type, and Fishbone-type—were evaluated using the material SCHULAMID® 6 MV14 FR4 K1681. The simulation results revealed that runner geometry has a decisive influence on filling uniformity, and they further demonstrated the effectiveness of a pressure-controlled runner approach in improving overall product quality. The findings highlight the importance of rheology-based optimization in runner-system design. This study differentiates itself from previous research by providing a comparative analysis of multiple runner types and by demonstrating that balanced filling can be successfully achieved not only in molds with 2ⁿ cavity counts but also in intermediate cavity numbers such as 12 and 14. The rheology-based pressure-controlled methodology presented here introduces a new optimization perspective for multi-cavity injection mold design.

Anomaly Detection System for Distributed Job Processing within Microservice Architectures

Ramazan Pekin, Kerem Bozkurt — Sun, 21 Dec 2025 00:00:00 +0300

Mobile payment systems process millions of transactions daily across distributed microservice architectures, where operational anomalies and silent failures can lead to financial losses and system instability. Traditional threshold-based monitoring is insufficient for detecting subtle, context-dependent deviations that evolve with user behavior and workload patterns. This study introduces a self-learning hybrid anomaly detection framework that integrates Isolation Forest, LSTM Autoencoder, and One-Class SVM to capture statistical, temporal, and structural deviations in operational metrics. Model outputs are fused using a calibrated soft majority voting strategy based on normalized anomaly scores. The trained framework is deployed as a containerized microservice, enabling real-time anomaly assessment based on live operational statistics. Experimental evaluation across a fifteen-month dataset demonstrates that the ensemble improves detection robustness and reduces false negatives compared to individual models and simple averaging strategies. The results highlight the system’s ability to detect silent failures and abnormal behaviors that occur without explicit exceptions while maintaining scalability and adaptability in complex financial microservice environments.

AI-Powered Multi-Agent Fashion Assistant for Personalized Retail Recommendations

Seza Dursun, Sedat Çelik, Bahar Önel, Tülin Işıkkent, Mert Alacan — Wed, 24 Dec 2025 00:00:00 +0300

As fashion retail navigates a new era shaped by heightened consumer expectations and rapidly evolving digital interactions, the need for deeply personalized, stylistically coherent, and context-aware recommendation systems has become paramount. Traditional engines, reliant on static rules or collaborative filtering, often fall short in capturing the complexity of human taste and the visual-semantic richness inherent in fashion products. This paper introduces Boyner’s AI-powered Multi-Agent Fashion Assistant, an enterprise-grade personalization platform architected on Microsoft Azure AI Foundry. The system orchestrates multiple specialized agents to deliver real-time, occasion-aware, and visually grounded fashion recommendations across omnichannel touchpoints. Leveraging multimodal embeddings, behavioral clustering, semantic search, and real-time trend signals, each agent operates with a distinct cognitive function, from silhouette-based outfit pairing to brand–season compatibility evaluation. Our implementation demonstrates how agentic AI systems can bridge the gap between algorithmic precision and stylistic intuition in large-scale fashion environments. The assistant not only enhances conversion and engagement metrics but also redefines the digital shopping journey as an explainable, adaptive, and human-centric dialogue. By operationalizing multi-agent orchestration within a live retail environment, Boyner pioneers a new paradigm in AI-powered visual discovery, offering a scalable blueprint for next-generation personalization in the global fashion ecosystem.

A Web-Based Credit Card Payment Architecture for Dealer Portals: Android POS Integration, Microservice Design, and Behavioural Segmentation for Data-Driven Dealer Management

Adnan Erdogan, Hüseyin Oktay Altun — Wed, 24 Dec 2025 00:00:00 +0300

Digital transformation in financial services has accelerated the need for secure, scalable, and user-centric payment infrastructures across various industries This study presents the design and implementation of a web-based credit card payment architecture integrated into the Dealer Web Portal (BWP), enabling dealer-initiated bill payments through Android POS ecosystem. The work covers three major dimensions: the development of a microservice-based web architecture using REST/SOAP services; real-time, bi-directional communication between the web portal and Android POS devices; and an unsupervised machine learning framework for behavioural segmentation using large-scale bill payment data. Multiple clustering algorithms, including K-Means, DBSCAN, Mean Shift, Spectral Clustering, and Hierarchical Clustering, were evaluated, with K-Means yielding the most meaningful segmentation results based on Purity, NMI, and Silhouette metrics. Segment outputs enabled dynamic commission policies, targeted dealer interventions, and time-series behavioral insights. The results demonstrate that the proposed architecture significantly enhances operational efficiency and data-driven decision making. This study provides one of the first integrated examples of Android POS–web portal interoperability combined with large-scale behavioural segmentation in Türkiye’s bill-payment ecosystem.

Investigation of the Comfort and Quality Properties of Knitted Garments Produced with Raised Yarn

Fri, 26 Dec 2025 00:00:00 +0300

The raising process is a finishing treatment applied to textile surfaces to impart softness and bulkiness. Traditionally performed on fabrics, this process creates a fluffy structure by pulling fiber ends to the surface, resulting in a fuller handle and enhanced comfort. In recent years, the adaptation of this technique to yarn form, known as the “yarn-level raising process,” has emerged as an innovative approach in textile manufacturing. When applied to yarns, the process generates a micro-level hairiness on fiber surfaces, increasing yarn volume and heat retention capacity. Consequently, fabrics knitted from such yarns exhibit higher air-holding capacity, lower thermal conductivity, and improved moisture management, leading to enhanced tactile softness and overall thermal comfort.

This study aims to investigate the effects of the yarn-level raising process on the thermal comfort performance of knitted garments. In today’s textile industry, the demand for comfort-oriented products—particularly those providing thermal comfort—has been steadily increasing. Although knitted fabrics offer advantages such as flexibility, lightness, and softness, their thermal insulation and moisture transfer capacities remain limited and require improvement.

Within the scope of this research, process parameters including fiber type, fiber fineness (nm), and feed rate (m/min) will be examined. Their influence on key performance criteria such as thermal conductivity, air permeability, and moisture transfer will be experimentally analyzed. All tests will be conducted using internationally recognized standard methods.

The findings are expected to contribute to the development of comfort- and quality-oriented process strategies in knitted garment production and to scientifically demonstrate the potential of yarn-level raising as an effective method to enhance thermal comfort and moisture management in textile materials.

Development of a Process to Prevent Back Contamination Caused by Cationization After Cationic Digital Reactive Printing on Cotton Knitted Fabrics

Dervis Malatyali, Selcen Iremnur Baloglu, Sena Efsun Alpaslan — Thu, 27 Nov 2025 00:00:00 +0300

Due to the restricted fixation and hydrolysis of reactive dyes, digital inkjet printing on cotton materials presents difficulties with low color output and substantial wastewater formation. Cotton can be cationically modified to improve color strength, decrease salt requirements, and boost dye absorption and fixation. However, conventional two-stage cationization methods are time-consuming and water-intensive, and they frequently result in back staining when the cloth is washed, with unfixed colors discoloring the white (unprinted) portions. By adopting a rotary printing process to directly put cationic printing primer onto cotton knitted fabrics, this study explores a novel one-step method to address these problems. We created three distinct pretreatment formulations: two with varying quantities of sodium hydroxide and 3-chloro-2-hydroxypropyl trimethylammonium chloride (CHPTAC) and a reference with no cationic ingredients. These formulations were applied to fabric, then digitally printed, air dried and steamed. In order to assess how well the printed fabrics prevented back-contamination, they were subsequently put through two distinct post-washing techniques; rope washing and open-width washing. The primary objective was to determine whether a combined cationization and printing process could simplify workflow and significantly reduce water and chemical consumption while ensuring print quality. The level of back contamination was assessed qualitatively by visually assessing the contamination of white areas after each washing process. The results from this study will provide important insights into the discussion on the industrial applicability of cationic cotton, particularly by addressing the issue of persistent contamination and exploring more sustainable, one-step processing solutions. The results obtained from this study contribute to the development of more efficient and environmentally friendly digital printing processes for cotton fabrics by providing important insights into the discussions on the industrial applicability of cationic cotton fabric, particularly by addressing the issue of persistent contamination in cationization and investigating more sustainable, one-step process solutions.

Environmentally Friendly Textile Printing Methods: Examining Natural Dyes as Alternatives

Merve Yarali Kinli; Ceren Göde, Süleyman Ilker Ertuna, Serkan Alsan — Wed, 27 Aug 2025 00:00:00 +0300

The textile industry has significant environmental, economic, and social impacts, extending beyond production to marketing and consumption. High energy and water usage, synthetic chemical pollution, and the fast fashion model contribute to ecosystem disruption, resource depletion, and hazardous waste generation. Sustainable transformation in this sector is critical, particularly in dyeing and printing processes, which are among the most environmentally damaging stages. Natural dyes present a promising alternative due to their biodegradability and lower ecological footprint. However, technical limitations such as low colorfastness and inconsistent results hinder their industrial adoption. This study investigates the performance of algae-based natural dyes developed using Algaeing technology. Algae dyes offer high colorfastness, seasonal color consistency, and significantly reduce environmental impacts—cutting greenhouse gas emissions by up to 70% and water usage by up to 98%. Free from toxic substances, they promote a safer, more sustainable textile dyeing process. The research highlights the potential of algae dyes as a viable solution for improving sustainability in the textile industry.

DroidDissection: A Hybrid Analysis Framework for Android Malware Detection and Analysis

ilker Kara — Wed, 17 Sep 2025 00:00:00 +0300

The Android operating system dominates the mobile ecosystem due to its flexibility, large application market, and open-source architecture. However, these same characteristics make Android an attractive platform for attackers who distribute malicious applications, particularly those designed to intercept banking transactions and steal confidential information. Existing security mechanisms mostly rely on either static or dynamic inspection, and these isolated techniques often fail to reveal concealed or runtime-triggered malicious behavior.

In this study, we present DroidDissection, a framework designed specifically for Android malware detection with an emphasis on banking-related threats. The framework combines static code and permission inspection with controlled dynamic execution, enabling deeper observation of behavior that only emerges during runtime. A real malware sample was examined to validate the approach. The experimental results show that the hybrid inspection strategy increases the accuracy of malware identification and helps uncover behaviors that traditional individual methods may overlook. These findings indicate that the proposed framework can strengthen defense mechanisms against evolving cyber threats targeting Android devices.

An Empirical Comparison of Claude, Llama, and Gemini for Aspect-Level Sentiment

Pınar, Mustafa — Sun, 23 Nov 2025 00:00:00 +0300

Aspect-based sentiment analysis provides granular insights into customer feedback by identifying discrete aspects, such as features or topics, and assigning a corresponding sentiment to each. This study assesses three large language models, hereafter referred to as LLMs, namely Google Gemini 2.5 Flash-Lite, Anthropic Claude Sonnet-4 delivered through AWS Bedrock, and Meta LLaMA 3.3 70B delivered through AWS Bedrock, using a real-world multilingual corpus of 7,841 Turkish mobile banking app reviews from İşbank in Turkey. We employ a prompt-based tagging protocol to extract aspect–sentiment pairs from every review, and we compare accuracy, F1-score, inference cost, and latency. The results show that all three LLMs can execute multilingual aspect extraction and sentiment categorization without task-specific fine-tuning. Claude Sonnet-4 attains the highest F1 for aspect extraction and the highest sentiment accuracy, although it incurs a markedly higher inference cost. Gemini 2.5 Flash-Lite achieves competitive accuracy at a fraction of the price, making it well-suited for high-volume analytics. Meta LLaMA at the 70B scale accessed through AWS Bedrock exhibits intermediate performance with moderate cost and latency. We provide detailed performance tables and figures, along with best-practice guidance for enterprise deployment. AWS Bedrock enables the strategic selection of Claude and LLaMA 3.3 70B for multilingual sentiment analysis, offering valuable insights from app reviews within scale, accuracy, and budget constraints.

Evaluation of ROPS and FOPS Tests for Structural Integrity of Forklifts

Hüseyin Samet Kartal, Orkan Buran, Mustafa Demir, Ali Can Tellioğlu — Mon, 24 Nov 2025 00:00:00 +0300

This study examines the Roll-Over Protective Structure (ROPS) and Falling Object Protective Structure (FOPS) tests and their results for forklifts. The main focus of the study is the evaluation of the cabin safety of a forklift with a lifting capacity of 3.5 tons. Within this scope, the compliance of the protective structures with the international standards ISO 3471 and ISO 3449 has been thoroughly analysed.

In the ROPS tests, the applied force and energy values under lateral, rear, and vertical loading conditions were calculated; the test setup, the applied loads, and the resulting deformations in the structure were investigated. Furthermore, in the FOPS test, the required energy levels were evaluated, and the results of the tensile tests conducted to verify the mechanical integrity of the fasteners (bolts) after the impact test were presented.

A Decision Support Framework for Customer Loyalty Program Managers: Reward Mix Optimization

Ayşe Salı, Sahika Koyun Yilmaz, Meryem Ezgi Aslan, Gizem Temelcan Ergenecoşar — Thu, 27 Nov 2025 00:00:00 +0300

Customer Loyalty Programs are a proven methodology for establishing and maintaining customer relationships. With the development of mobile technologies and the power of digitalization, what was once a simple punch card has now evolved into a full-fledged mobile application. The paradigm shift has opened up research areas on an individual customer level, especially in non-contractual traditional commerce, which was previously impossible due to a lack of loyalty data. The cost and budget of Customer Loyalty Programs increase with their strategic value. Balancing the attractiveness of a reward to the customer with the unit cost to the organization is essential for designing effective programs. In this study, we propose a framework that combines the attractiveness and unit cost of rewards to provide an optimized reward mix, thereby aiding Customer Loyalty Program managers in their decision-making processes.

An Innovative Approach to Technical Textiles: Assessing the Performance of Olefin-Based Outdoor Fabrics

Merve Yarali Kinli, Ceren Göde, Süleyman Ilker Ertuna, Serkan Alsan — Wed, 03 Dec 2025 00:00:00 +0300

In recent years, consumer preferences have undergone significant transformations driven by socio-cultural, economic, and demographic factors, resulting in a growing demand for innovative and sustainable textile products. The intensifying competition in the global market, along with evolving consumer expectations, necessitates that firms within the textile industry not only adapt to technological advancements but also develop sustainable solutions. Within this framework, Menderes Tekstil has emerged as a pioneer in the field of outdoor technical textiles by introducing an innovative production process through the utilization of olefin yarns. Compared to acrylic and polyester yarns, olefin-based fabrics demonstrate superior mechanical performance, offering twice the tensile strength of acrylic and approximately 30% greater strength than polyester. In addition, these fabrics exhibit water- and oil-repellency, ultraviolet resistance, mold resistance, and stain resistance, thereby ensuring durability and suitability for outdoor applications. The recyclability of olefin yarns further reinforces an environmentally responsible production approach, contributing to broader sustainability objectives. Specifically developed for demanding applications such as garden furniture, umbrellas, awnings, and marine textiles, these products not only address a critical gap in the textile market but also enhance the competitive capacity of the firm by combining innovative features, environmental advantages, and high added value.

Development of Ash-Based Paving Stones Through the Utilization of Industrial Ash Generated During Urban Waste Disposal Processes

Cihangir Cebeci, Mahmut Uğur, Ece Yaralı, Gözde Bostancı — Fri, 12 Dec 2025 00:00:00 +0300

With the increase in urban waste and industrial activities, significant quantities of ash are generated from waste incineration plants, cement factories, and energy production facilities. These ash wastes pose critical environmental challenges such as land occupation, soil contamination, and groundwater pollution, making their sustainable management essential. This study aims to utilize industrial ash waste as an alternative raw material in the production of paving stones used in infrastructure applications. Ash-based paving solutions reduce dependence on natural stone and sand resources, contributing to resource conservation; additionally, they help lower environmental impacts by reducing cement consumption and related carbon emissions.

Within the scope of this study, manufacturability, mechanical strength, durability, and environmental performance criteria were evaluated. The results indicate that ash-based paving blocks offer an economical and sustainable alternative to conventional products. This approach enables the transformation of waste materials into value-added products instead of disposal, supporting the development of low-cost and environmentally friendly building materials.

Aligned with European Union environmental policies, this study supports the use of long-lasting, environmentally conscious, and aesthetically favorable products in urban planning and infrastructure applications. The findings demonstrate that industrial ash waste can become a valuable resource for the construction materials industry.

Structural Behavior Analysis of Rail-Mounted Portal Cranes Equipped with a 360° Rotatable Spreader Mechanism Using the Finite Element Method

Samet Dönerkaya, Kemalettin Kök, Muhammed Emin Tamer — Wed, 10 Dec 2025 00:00:00 +0300

Rail-Mounted Gantry (RMG) cranes are complex lifting systems widely used in container terminals and industrial sites to ensure the safe handling of heavy loads in both horizontal and vertical planes. One of the main subsystems of these cranes is the trolley , which operates in conjunction with the spreader responsible for carrying and transferring the load. In recent years, designs incorporating 360° rotatable spreader mechanisms have provided significant flexibility in load positioning but have also introduced complex stress distributions in connection regions. In this study, the structural behavior of the spreader component of an RMG-type portal crane was analyzed using the Finite Element Method (FEM). The investigation covered static, buckling, and fatigue strength assessments, with all calculations performed in accordance with DIN EN 13001-3-1+A2 and EN 13001-3-8 standards. During the modeling phase, the steel framework of the spreader structure, along with its welded and bolted joints, was represented in detail. Separate loading scenarios were established for different spreader configurations (rotated positions of 0°, 45°, and 90°), and lifting loads, wheel reaction forces, and boundary conditions were applied in compliance with relevant standards. The analysis results showed that the maximum Von Mises stresses remained below the material yield limit, and the fatigue strength in critical connection areas (such as drum plates, connecting bolts, and weld seams) satisfied the S6 stress range class requirements. Furthermore, the deformations observed in the spreader were within allowable limits, confirming that the structure possessed adequate rigidity against buckling. In conclusion, the RMG portal crane design equipped with a 360° rotatable spreader system was found to be structurally safe in terms of both static and fatigue performance, demonstrating compliance with international standards. This study contributes to the engineering-level optimization of spreader–trolley interaction in next-generation RMG crane systems.

Investigation of Mechanical Properties of Hemp Hurd/PP Composites for the Application of Water Irrigation Pipes

Özgür Demircan, Hüsnü Armağan Gümüş, Murat Kuru, Beyza Gizem Duman — Tue, 09 Dec 2025 00:00:00 +0300

In this study, for the first time, a new generation hemp hurd/PP composite material coupling sleeve prototype production was performed for the water irrigation pipes. Within the scope of experiments, at first hemp hurds were prepared by using cyclic grinding machine. Later, the compounds of hemp hurds (0 wt%, 10 wt%, 20 wt% and 30 wt%) and polypropylene (PP) were prepared using double screw extruder machine at Ondokuz Mayıs University. After that, the specimens of 0 wt%, 10 wt%, 20 wt% and 30 wt% hemp hurd reinforced polypropylene (PP) composites were fabricated using injection molding machine. Three points bending tests were performed on the fabricated specimens with INSTRON 5982 100 KN universal test device at Ondokuz Mayıs University (OMU) KITAM Central laboratory. Prototyping of hemp hurd/PP composite material coupling sleeves were produced using plastic injection machines of Poelsan Plastik Sanayi ve Ticaret A.Ş. Long-term tightness test under internal pressure was conducted on the fabricated coupling sleeves in Poelsan Plastik Sanayi ve Ticaret A.Ş. According to the bending test results, the bending modulus of specimens were increased by increasing hemp hurd content. The highest bending strength was obtained by 10 wt% hemp hurd powder reinforced PP composites (46.5 MPa). The findings showed that the coupling sleeves manufactured from hemp hurd/PP composite material can be successfully used as an alternative to %100 PP material coupling sleeve under similar service conditions in water irritation systems.

Artificial Intelligence-Assisted Control of Light Pipe & LED Luminaire Hybrid Tunnel Lighting System

Levent Doğan, İsmail Kıyak — Tue, 23 Dec 2025 00:00:00 +0300

Tunnels are designed as infrastructure elements that facilitate smoother traffic movement, enhance operational safety, and minimize environmental effects. However, when adequate lighting is not provided in tunnels, sudden transitions from bright outdoor environments to dim indoor spaces cause temporary vision loss while the eyes adapt to the new environment. Sudden changes in light at tunnel entrances and exits can disorient drivers and increase accident risks. Daylight offers a mix of wavelengths and color temperatures that provide optimal visual conditions for humans. In this study, an energy-efficient hybrid tunnel lighting system combining light tubes with artificial lighting was designed, and an artificial intelligence–based control system dependent on daylight was developed for this setup. To make tunnel conditions more efficient and comfortable for drivers, a control system incorporating an artificial neural network (ANN) algorithm was designed to apply the instantaneous outdoor illuminance level at the tunnel entrance. The control system results were analyzed, indicating that approximately 25.30% energy savings can be achieved compared to conventional lighting control methods, along with an expected improvement in drivers’ visual comfort.

Designing for Explainability and Data Sovereignty: A Design Principles Approach for LLM-Augmented FinTech Analytics

Begüm Al, Adnan Erdoğan, Oğuzhan Akkurt, Nazım Taşkın, Yavuz Acar — Thu, 18 Dec 2025 00:00:00 +0300

This study reports on the design and development of a practical analytics system that responds to the growing need for data-driven work among users without formal training in programming or data science. The system uses large language models (LLMs) to support natural language interaction and to guide users through common data analysis tasks. Compared with typical analytics tools, the system does more than simply run models in the background. It explains in plain language what each model is doing and why particular results appear, and it walks the user through the choice of methods step by step. The architecture can connect to different locally deployed LLMs – for example LLaMA, Qwen or DeepSeek, so organisations are not locked into a single provider. All interaction takes place through a chat-style interface: users upload a dataset, describe their question, and the system handles the configuration and code. The artefact was shaped through a Design Science Research (DSR) process, with several iterations of design, feedback and revision involving potential users. In its current form, a proof-of-concept implementation and scenario-based examples show that non-technical users are able to understand their data more clearly and make more informed choices among analytical options. Taken together, these features point to a practical and adaptable framework that brings explainable, LLM-supported analytics within reach of a much wider group of professionals.

Classifying Operator Experience from Electric Screwdriving Signals: A BiLSTM-Based Study with External Validation

Kader Nikbay Oylum, Turgay Tugay Bilgin — Wed, 26 Nov 2025 00:00:00 +0300

This study presents a deep learning–based approach to objectively classify operator experience levels (Novice–Intermediate–Expert) from multivariate signals and user interactions obtained during electric screwdriving operations. The dataset comprises 64 participant-specific files, each containing multiple tightening trials. Windowing was performed independently per file; short segments unsuitable for windowing were excluded, yielding 3,326 time windows (2,958 for training/testing; 368 for independent validation). A two-layer Bidirectional LSTM (BiLSTM) architecture was employed and evaluated on both the train–test split and an external validation set constructed from 12 previously unseen files. On the test set, the model achieved 76% overall accuracy with macro-averaged precision/recall/F1 of 77%/76%/76%. Class-wise analysis indicated stronger separability for the Expert class (recall ≈ 84%) and comparatively lower performance for Intermediate (recall ≈ 66%). On the hold-out validation set, accuracy was 75.00%, with a mean predicted probability of 85.0%, indicating moderate-to-high confidence. The findings show that while BiLSTM provides a solid foundation for time-series classification, its effectiveness may be limited for complex patterns without a convolutional front end.

Optimization of Pultrusion Process Parameters for Carbon Fiber/Epoxy Composites

Sibel Demir, Ömür Alkan — Thu, 27 Nov 2025 00:00:00 +0300

This study investigates the effects of key pultrusion process parameters—including temperature profile, fiber volume ratio (FVR), preformer geometry, resin viscosity, and line speed—on the production stability and mechanical performance of carbon fiber/epoxy composite profiles. Continuous carbon fiber rovings were impregnated with epoxy resin and processed through a multi-zone heated die under varying operating conditions. Tensile properties were evaluated in accordance with ASTM D3039 to ensure standardized and comparable mechanical characterization. Experimental observations revealed that even small adjustments in thermal management, heating zone positioning, preformer compression and eye diameter, fiber volume ratio, resin rheology, fiber type, squeezer configuration, and pulling speed produced significant variations in surface quality, flow behavior, resin backflow, fiber congestion, and overall process stability. The optimal process window was achieved at a line speed of 30–35 cm/min and an FVR range of 65–70%, with improved results obtained by shifting the initial heating zone backward, reducing the final preformer diameter, and utilizing lower-viscosity resin systems. The findings provide a comprehensive process–property relationship for carbon pultrusion and offer a practical guideline for industrial optimization aimed at achieving stable production and high-quality composite profiles.

Improving the Accuracy of Location Data in UWB-Based RTLS Using Deep Learning Methods

Ramazan Kavak, Fatih Aydemir, Serap Cekli — Fri, 28 Nov 2025 00:00:00 +0300

In Real-Time Location Systems (RTLS) using Ultra-Wideband (UWB) technology, the Decawave DW1000 uses the Two-Way Ranging (TWR) method to obtain the location of a moving object. Multipath propagation occurring under NLOS conditions systematically negatively affects time-leads and distance measurements; this increases the bias (positive bias) and widens the variance, leading to instability in the location data. In this study, an autoencoder-based measurement improvement method proposed for the tag location data obtained using the TWR method. The raw TOF (time of flight) and range measurements obtained from the DW1000 are simultaneously integrated into a low-dimensional latent space with features such as RSSI and CIR-based quality metrics (e.g., first-path amplitude/index, channel energy, pulse width indicators). The denoising/regularized reconstruction process suppresses the jump and bias components in the location data caused by NLOS; thus, the improved measurements can increase the stability and repeatability of location data when used with classical Gauss-Newton location. This approach can trained with a highly dynamic setup (especially using clean LOS records), reducing the burden of relying on field geometry; its modular architecture allows for minimal integration into the existing TWR software chain. Experimental analysis and visualizations were performed on different indoor scenarios (office, corridor, and semi-open space layouts) using MATLAB. This method has been shown to provide a consistent reduction in mean error metrics (MAE/RMSE), a significant improvement in axis bias errors (95th/97th percentile), and location path continuity, while also eliminating erroneous outliers originating from instantaneous NLOS

Impact of Phase Inversion and Process Parameters on Alkyd Emulsion Properties

Pelin Aksoy, Selim Çifçi — Fri, 23 May 2025 00:00:00 +0300

In conventional alkyd coatings, the alkyd resin is dissolved in organic solvents, such as mineral spirits. However, due to growing health and environmental concerns, there is a rising interest in waterborne alkyd coatings. Alkyd emulsions, therefore, represent a viable alternative. Waterborne alkyd resins significantly reduce VOC emissions by using water as the dispersion medium, making them highly important for research in this area. The main objective of this study is to synthesize emulsified alkyd resins by replacing solvents with water, using nonionic and anionic surfactants, for the alkyd resins widely used in the paint industry for many years. Soybean oil/sunflower oil alkyd resins were emulsified with the anionic surfactant MAXEMUL 7201 and the nonionic MAXEMUL 7101. This paper investigates the effect of emulsification temperature, mixer model and rotation speed on the formation of alkyd emulsion The synthesized alkyd resins were evaluated based on acid value, viscosity, gloss, Fourier transform infrared spectroscopy (FTIR), gel permeation chromatography (GPC), and particle size analysis.

Machine Learning Models and Explainable Artificial Intelligence Approaches for Intrusion Detection in IoT Networks

Asuman Besi Kütük, Özlem Çoşkun, Hikmet Kütük, İbrahim Kök — Sat, 31 May 2025 00:00:00 +0300

The rapid spread of Internet of Things (IoT) technologies and the rapidly increasing use of IoT devices offer technological transformation and innovative solutions in many areas from daily life to industrial processes. However, the resource constraints, simple operating systems, non-standard protocols and embedded software of IoT devices make them vulnerable to cyber-attacks. This makes IoT networks risky against malicious attacks and increases the size of security threats. Moreover, the complexity and heterogeneity of IoT networks render traditional security approaches inadequate and increase the need for advanced solutions. In this context, the need for methods for detecting and preventing attacks on IoT networks that are not only reliable and effective, but also understandable by users and security experts has become increasingly critical. This need for network security necessitates the development of strategies that will both secure technical infrastructures and increase the trust of human elements interacting with these infrastructures. In this context, the need for more interpretable, explainable and transparent security approaches is increasing. In particular, machine learning (ML) and deep learning (DL) based intrusion detection systems offer effective solutions to security problems such as anomaly detection and attack classification. The comprehensibility of the decision mechanisms of the models used enables both security experts to manage the systems more effectively and users to have more confidence in the security measures taken. Explainable Artificial Intelligence (XAI) techniques make the decision processes of ML and DL models transparent, allowing to understand how and why attacks are detected. Accordingly, it has become a critical requirement for security systems not only to achieve high accuracy rates, but also to make the decisions taken interpretable. In this study, the effectiveness of artificial intelligence (ML and DL) techniques for the detection and classification of security threats in IoT networks is analysed. In addition, the applications of XAI methods such as SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME) and Explain Like I'm 5 (ELI5) for IoT security are investigated. It is shown how these methods make the decision processes of ML and DL models used in IoT networks more transparent and provide a better analysis. As a result, this study presents an approach that combines both performance and explainability in IoT security. By demonstrating the effectiveness of XAI-supported ML and DL models, it aims to contribute to future research and innovative security solutions for enhancing security in IoT networks.

Two-Factor Authentication System for Protecting Metadata and Connected Vehicles

Huseyin Karacali, Nevzat Donum, Efecan Cebel — Fri, 27 Jun 2025 00:00:00 +0300

Connected vehicles are becoming increasingly common, and they store a significant amount of data about their drivers and their surroundings. This data is attractive to attackers, and there is a need for effective security measures to protect it. This study proposes a two-factor authentication (2FA) system to protect the metadata stored in connected vehicles and the vehicles themselves. The system consists of two main components: the Central Security Unit (CSU) and the AutoGuard (AG) mobile application. The CSU is integrated with the Remote Keyless Entry System (RKES), while the AG is installed on the authorized driver's phone. The 2FA process begins when the remote key is in proximity to the vehicle. This triggers the CSU, which then initiates the second authentication factor. The AG prompts the driver to enter a valid security method, such as a biometric, pattern, or PIN code. If the second authentication is successful, the AG authorizes the CSU and the vehicle doors are opened by the RKES. The driver is notified by CSU through AG if the 2FA process is unsuccessful. As a result, this system aims to protect the metadata stored in authorized users' vehicles and their vehicles from unauthorized invaders. Furthermore, it is possible to enhance the accuracy of the 2FA system by integrating the location of the phone with AG functionality into the authentication system.

A Review of Deep Learning Approaches with CMR Images in the Diagnosis of Cardiovascular Diseases

Gülsüm Kemerli, Tayyip Özcan — Fri, 27 Jun 2025 00:00:00 +0300

Cardiovascular disease (CVD) is one of the leading causes of death worldwide, which has led to the recent intensification of Deep Learning (DL) studies in the field of cardiology. Patients usually experience symptoms such as rapid fatigue, edema below the knee and ankle, chest pain, shortness of breath, and palpitations. The most common types of CVD include coronary artery disease, arrhythmias, congenital heart defects, cardiomyopathy, valvular heart failure, and angina. Electrocardiography (ECG), blood tests, physical examination, and medical imaging are the most effective tools for diagnosing diseases. In recent years, cardiac magnetic resonance imaging (CMRI) has been increasingly used for the diagnosis, follow-up, treatment planning, and prognosis of CVDs. However, the large number of slices and low contrast of CMRI data make diagnosing CVD difficult. Deep learning techniques are being applied to diagnose CVD with CMRI data to overcome these difficulties, and intensive research continues to be conducted in this field. It is important to keep abreast of developments so that these studies can significantly impact clinical applications. This review aims to be a stepping stone for researchers in this process by comprehensively reviewing studies on CVD detection using DL methods on CMRI images.

A Data Fusion Method Combining Image, Sensor, and Survey Data for Efficiency and Usability Analysis of Electric Power Tools in Industrial Environments

Kader Nikbay Oylum, Turgay Tugay Bilgin — Sat, 28 Jun 2025 00:00:00 +0300

The increasing integration of advanced technologies and automation in industrial production has heightened the importance of operational efficiency and safety. Among the critical components influencing workforce performance and product quality is the effective use of electric hand tools. However, the limited availability of comprehensive datasets and the absence of robust labeling methodologies present significant challenges for accurate data analysis and predictive modeling. This study addresses these limitations by incorporating field-collected data and multiple data acquisition techniques to identify relevant features for machine learning applications. An initial dataset comprising 51 attributes was systematically reduced to 16 through feature selection processes, enhancing its suitability for subsequent computational modeling. Several classification algorithms were evaluated for data labeling, with the Decision Tree method demonstrating superior performance in terms of accuracy. Despite these promising results, the dataset’s limited sample size (64 individuals) restricts the generalizability and reliability of machine learning outcomes. To mitigate this constraint, data augmentation techniques will be employed to generate synthetic instances, thereby expanding the dataset. Upon achieving a sufficient sample size, machine learning models will be developed to predict individuals’ proficiency with electric hand tools. This research contributes to the foundational knowledge required for efficient data collection, accurate labeling, and the development of predictive models in industrial settings.

Mechanical and Environmental Comparison of Natural Fibers and Glass Fiber in the L-RTM Method

Nazmi Türkhan, Kübra Tuna, Hüseyin Arıkan, Yusuf Uzun — Sun, 29 Jun 2025 00:00:00 +0300

The applicability of flax fiber-reinforced composites as an environmentally friendly alternative to glass fiber-reinforced plastics (GFRP), commonly used in outdoor structures such as water slides, has been investigated. While glass fiber is associated with high energy consumption and significant environmental impacts, flax fiber offers a sustainable solution due to its renewable nature, low density, and biodegradable properties. The mechanical and environmental performance of flax fiber-reinforced composites manufactured using the L-RTM (Light Resin Transfer Molding) method was evaluated, with a particular focus on sensitivity to water and moisture, and design considerations to mitigate these effects were discussed. In this method, L-RTM is employed as a vacuum-assisted, closed-mold technique particularly suited for medium-scale production of high-quality components with smooth surfaces on both sides. In this context, the essential conditions for natural fiber-reinforced composites to serve as a viable alternative to glass fiber in water slide applications have been identified.

Smart Classroom Scheduling and Resource Optimization for Educational Institutions: Integrating AI and Multi-Objective Decision Support

Osman Çaylı, Atınç Yılmaz — Mon, 07 Jul 2025 00:00:00 +0300

The demand for high-quality education delivery in increasingly dynamic and competitive educational markets has intensified the need for intelligent and adaptive scheduling systems. Manual classroom scheduling methods, which rely on human decision-making, often fail to optimize critical resources such as classrooms, teachers, and student time, leading to inefficiencies and economic losses. This paper proposes a comprehensive Smart Classroom Scheduling and Optimization System (LMSOPT) that leverages Artificial Intelligence (AI), advanced time-series forecasting, constraint-based multi-objective optimization, and real-time data integration. The proposed system employs Long Short-Term Memory (LSTM) neural networks for highly accurate demand forecasting, alongside heuristic and metaheuristic optimization algorithms such as Constraint Programming (CP), Genetic Algorithms (GA), and Tabu Search. The system aims to dynamically balance multiple conflicting objectives: maximizing classroom occupancy rates, minimizing student waiting times, and aligning teacher availability with student preferences. The expected contributions are multifold: significant operational cost savings, measurable improvements in resource utilization, increased student satisfaction, and the creation of an extensible research framework for AI applications in education management. The study aligns with national strategies for digital transformation and supports the vision of data-driven decision-making in educational administration. Empirical results and comparative analyses are presented to validate the system’s effectiveness and demonstrate its replicability for institutions of various scales.