Orclever Proceedings of Research and Development

An Integrated Deep Learning Framework for Automated Quality Control and Process Optimization in Slasher Indigo Dyeing

2026-01-12T11:49:35+03:00

This paper presents the development of a multi-step, multi-disciplinary automation framework designed to enhance quality assurance and process control in slasher indigo dyeing machines. The system integrates two complementary subsystems: (1) a real-time yarn defect detection module employing deep learning-based computer vision, and (2) a process optimization module utilizing chromaticity analysis for colour stability and chemical balance control. The defect detection system uses four moving cameras strategically placed across the machine to identify broken yarns and irregular density patterns with high accuracy. The colour monitoring subsystem, developed in collaboration with Agteks, continuously records yarn colour in the CIELAB colour space and recommends corrective pH or reduction agent (Hydro) adjustments when deviations occur. Experimental results demonstrate a detection accuracy of 92.4%, with significant improvements in production speed, consistency, and operator workload reduction. The proposed system represents a comprehensive step toward fully autonomous dyeing operations aligned with Industry 4.0 objectives.

Development of a Symbiotic Snacks Bar Product

2026-01-12T20:02:08+03:00

In this study, control (not containing inulin and probiotic) prebiotic and symbiotic (including probiotic and prebiotic) bar formulations were developed using whey protein concentrate (WPC), oat flour, inulin, or Lactobacillus rhamnosus (L. rhamnossus) GG. The total solids and protein values of the products were varied from 86.90 to 89.21% and from 20.85 to21.78%, respectively. The probiotic viable count in the symbiotic bar sample was 8.05 log cfu/g. Sensory analysis revealed that all samples exhibited similar visual characteristics, while the prebiotic bar received the highest taste and overall approval scores. The findings demonstrated that WPC and inulin were effective in preserving probiotic viability and product quality. In conclusion, the developed symbiotic bar formulation is considered an innovative product suitable for the healthy snack market, with both functional and nutritional properties.

Digital Balance in Agriculture

2026-01-12T11:49:03+03:00

This study introduces an electronically and mechanically controlled force balancing system designed to ensure safe and stable operation of agricultural machinery on sloped terrain. Equipment attached to the rear of tractors and construction vehicles often disrupts the load distribution, particularly on inclines, increasing the risk of front-end lift and rollover. The proposed mechanism continuously monitors the dynamic load distribution via weight sensors and accelerometers integrated into the front axle, enabling real-time analysis and response. When the measured load falls below predefined thresholds, the system automatically adjusts the hydraulic tank volume to restore the vehicles center of gravity. This process is managed by an electronic control unit (ECU) and can be operated either automatically or manually, significantly enhancing vehicle stability and operational safety. The modular and vehicle-independent design allows seamless integration with various tractor and machinery models, while the multifunctional tank structure-serving as a tool compartment, auxiliary oil reservoir and fuel tank offers practical advantages under field conditions. This innovative approach provides a scalable, sustainable and high-safety solution to mechanical balance challenges in digitalized agricultural operations.

Chewable Tablet Containing Immune-Supportive Bioactive Compounds with Synergistic Effects: Ülker Everwell Force

2026-01-12T11:51:40+03:00

In recent years, the growing consumption of dietary supplements has accelerated research and development efforts aimed at supporting the immune system. In response to the increased demand that peaked during the pandemic, TÜBİTAK initiated studies focusing on the synergistic effects of green tea, blueberry, pomegranate peel, and propolis extracts, all of which are well-documented for their antibacterial, antiviral, and antioxidant properties. The oral mucosa, as a primary gateway to the body, plays a critical role in immune defense by acting as a barrier against microbial agents. Therefore, the developed product aims to strengthen the defense mechanism of the oral mucosa through an oral mucoadhesive system. The formulation of Ülker Everwell Force includes gum arabic, green tea extract, blueberry extract, pomegranate peel extract, propolis extract, vitamin C, and zinc. These standardized extracts are supported by literature for their antioxidant, antibacterial, and antiviral activities. According to the study conducted by Karaoğlu et al. on antiviral activity, the extracts exhibit a sevenfold increase in effectiveness when used in combination compared to their individual application. TÜBİTAK utilized this blend in a sorbitol-based compressed tablet and filed a European patent application titled “Production of Protective Lozenge/Chewable Tablet Against SARS-CoV-2 Virus” (EP23704823.6), which has received an intention-to-grant decision from the European Patent Office. In collaboration with CCC R&D Center and TÜBİTAK, the project further enhanced the mucoadhesive system by incorporating gum arabic and improved functionality through the addition of vitamin C and zinc. Unlike previous antiviral studies, this research focuses not only on the bioactive properties of Everwell Force but also on its formulation design, mucoadhesive mechanism, and stability performance as a functional food matrix.

Innovative Technological Strategies to Enhance Bioavailability in Germinated Grains

2026-01-12T11:49:09+03:00

This study investigates the enhancement of bioavailability in wheat, oats, and barley through germination combined with advanced technological treatments. The experimental design included three cereal species (Triticum aestivum, Avena sativa, Hordeum vulgare) subjected to germination for 24, 48, and 72 hours, followed by either ultrasonic (20 kHz, 30 min) or low-temperature microwave pretreatment (700 W, 3 min). Primary analyses included phytic acid content, total phenolic compounds, and antioxidant capacity. Results demonstrated that combining germination with technological treatments effectively improved nutrient bioavailability

Visual Discovery in Retail: Operationalizing AI-Powered Visual Search at Boyner

2026-01-12T11:49:11+03:00

In today's retail landscape, where millions of products and visual stimuli compete for customer attention, the integration of artificial intelligence into visual search has emerged as a crucial lever of operational efficiency. This paper presents Boyner Group's AI-powered visual discovery system, which enables customers to search using photos instead of keywords, making product discovery more intuitive and visually engaging. The architecture leverages a hybrid approach combining Large Language Models (LLMs), vision models such as GroundingDINO, and vector-based semantic similarity engines like SigLIP+Milvus to deliver scalable and high-accuracy image retrieval. The system, currently operational across the Boyner.com.tr ecosystem, supports enhanced filtering and storytelling capabilities, increasing customer satisfaction and conversion rates. The implementation process, system components, and operational results of this large-scale AI integration are explored, highlighting its transformative impact within omnichannel retail.

Keywords: Visual Search, Multimodal AI, GroundingDINO, SigLIP, Milvus, Retail Intelligence, Semantic Search, AI in E-Commerce, Omnichannel Retail, Customer Experience

The Development of a Platform as a Service for Game Key Distribution

2026-01-12T11:49:17+03:00

In the digital game industry, sales methods and revenue sources can vary. In recent years, distribution carried out through digital platforms has surpassed physical copy sales, significantly increasing its market share. In this dynamic industry, it is important for publishers to securely deliver game licenses belonging to end-users and to maximize stakeholder profitability in this process. Within the scope of this study, a digital game distribution platform has been developed that establishes seamless connections between content owners and sales channels, and efficiently distributes digital content procured from content owners to various sales channels. Through this platform, efforts have been made to strengthen the digital content ecosystem; studies have been conducted regarding product loading processes, seamless technical integration, optimization of promotional strategies, ensuring secure distribution, and increasing revenue streams.

Credit Scoring with Machine Learning Supported by E-Commerce Data

2026-01-12T11:51:45+03:00

With the rapid growth of e-commerce, the need for credit in e-commerce has increased. E-commerce platforms require high performance as a competitive advantage in their activities. Traditional credit risk models need improvement to sustain the performance expected by e-commerce platforms. In this study, we investigate alternative behavioral and transactional variables obtained from an e-commerce platform. We examine whether these variables improve the predictive performance of credit risk models beyond traditional financial data. Our research is based on a real e-commerce environment where a machine learning based credit scoring system was implemented. The study focuses on developing and evaluating a credit risk system that integrates platform specific behavioral data, such as shopping frequency, payment methods, Buy Now Pay Later (BNPL) repayment behavior, and wallet usage, with traditional financial and Credit Bureau(CB) indicators. Our findings demonstrate a significant improvement in model discrimination and Gini performance. The localized AI-driven credit scoring system achieved a low-cost, fast, and more accurate credit assessment.Order

The Green Step Upper: A Novel Sustainable Bonding Method Replacing Solvent-Based Adhesives in Footwear Upper Assembly

2026-01-12T11:51:23+03:00

The Green Step Upper project introduces an innovative, solvent-free adhesive application technology designed to transform stitching preparation processes in footwear manufacturing. Traditional methods rely on double-sided application of solvent-based adhesives, resulting in high VOC emissions, increased labor dependency, inconsistent quality, and environmental burdens. This project eliminates these limitations through a specially engineered single-surface adhesive tape system that provides stable fixation, homogeneous bonding distribution, and improved process efficiency. The research covers material selection, ergonomic apparatus design, prototype development, and performance testing across different upper materials. Results demonstrate a complete removal of solvent use, a significant reduction in operational time and labor costs, improved product consistency, and zero VOC emissions, enabling fully sustainable production. The proposed system offers a scalable and commercially viable model that can be adopted by both domestic and international manufacturers, contributing to national competitiveness while supporting global sustainability targets.

Natural Language Processing-Based Layered Reconciliation System for Financial Transaction Analysis

2026-01-12T11:49:34+03:00

With the widespread adoption of digital payment systems, the volume and diversity of financial transaction data have increased significantly. For payment institutions and electronic money companies in particular, the cross-verification of internal transaction logs with bank statements has become a critical requirement for ensuring financial security, accounting accuracy, and auditability. However, in practice, inconsistencies often occur between bank-side and firm-side records due to system interruptions, service errors, or manually entered transactions. This study presents a financial data reconciliation system based on Natural Language Processing (NLP) and rule-based analytical techniques, designed to detect inconsistencies by comparing bank transaction records with internal operational logs. The system, developed by Elekse, automatically retrieves millions of transaction records from multiple banks via the Finekra platform and classifies them by transaction type using key attributes such as description, date, amount, and IBAN. Throughout this process, NLP techniques are used to identify linguistic patterns, extract meaningful expressions, and assign the appropriate accounting codes through predefined rules, enabling the automatic reconciliation of records. As a result, the need for manual inspection is reduced, error detection is accelerated, and overall data accuracy is improved.

A Modular Semantic Kernel Agent for Automated Code Review and Refactoring Feedback

2026-01-12T11:49:12+03:00

In modern software development, maintaining clean, efficient, and reliable code is critical to team productivity and product quality. This paper introduces a modular Large Language Model (LLM)-based agent, designed using Microsoft’s Semantic Kernel framework, for automated code review and refactoring feedback. The agent leverages plugin-based function orchestration, Retrieval-Augmented Generation (RAG), and dynamic prompt engineering to analyze source code across multiple dimensions; including readability, efficiency, security, and adherence to best practices. Integrated into CI/CD pipelines and broader SDLC workflows, the system provides contextual insights, the system provides contextual insights, suggests specific improvements, and explains reasoning for each recommendation. Evaluation results across real-world open-source repositories demonstrate the agent’s effectiveness in reducing human review time while improving refactor quality. The modular design ensures adaptability to various programming languages and enterprise development environments. This research highlights the potential of agentic LLM systems to augment software engineering workflows with intelligent, transparent, and developer-aligned feedback mechanisms.

Keywords: Code Review, Semantic Kernel, Plugin Orchestration, Refactoring, Large Language Models, Agentic AI, Retrieval-Augmented Generation, Prompt Engineering

Airline Crew Hotel Assignment: An Optimization Framework for Fairness and Efficiency

2026-01-12T11:51:43+03:00

Crew fatigue management is a critical aspect of airline operations, and layover scheduling plays a key role in ensuring the well-being of flight crew members. The increasing complexity of airline networks and availability of multiple hotels at stations necessitates the development of a framework leveraging efficient assignment of crew to the hotels. Besides efficiency, there are pre-defined rules that should be considered such as group-based assignment and maximum quota. In order to efficiently handle the assignment process, this study proposes a novel optimization framework with Mixed Integer Programming (MIP) formulation that ensures compliance with the associated rules. Meanwhile, the proposed optimization model prioritizes fairness among the hotels at the same station. To validate the effectiveness of the proposed approach, an extensive set of experiments are conducted using real-world airline data and the results are analyzed thoroughly. The experiments depict that the proposed optimization model can be solved efficiently within a short time frame, making it suitable for large-scale airline operations.

Graph-Based Customer Segmentation with GraphSAGE on a Customer–Vehicle Bipartite Network

2026-01-12T11:49:37+03:00

This study models customer–vehicle interactions in an online used-car platform as a bipartite structure, constructing a graph with customer (U) and vehicle (V) nodes. Relations between the two node sets are defined only by edges representing realized purchase events (e=(u,v,t)), thereby focusing on a signal with high business value and relatively low noise. On this graph, inductive node representations (embeddings) are learned with GraphSAGE. During training, link prediction is used solely as a self-supervised proxy task; optimization employs an MLP-based scorer with Binary Cross-Entropy (BCE) loss. Early stopping is triggered when the BCE on a temporally held-out validation set stops improving; together with temporal negative sampling, this prevents leakage of future information.

The objective is to obtain high-quality customer/vehicle embeddings. The learned representations are then used to construct embedding-based customer segments via K-Means. Segmentation quality is evaluated using the Silhouette and Calinski–Harabasz scores. The results show that GraphSAGE embeddings learned on the purchase-induced bipartite graph provide a practical and scalable foundation for recommendation/targeting and customer understanding tasks

Development of a Secure Structural Component to Mitigate Environmental Contamination at Ports During the Transfer of Granular Materials in Global Maritime Logistics: Ecological Port Loading Bunker

2026-01-12T11:49:36+03:00

This study focuses on the design and production of the 100-m³ "Ecological Port Loading Bunker," which has been built for the first time in our country. The objective is to address the environmental pollution issue stemming from bulk material logistics in ports, which serve as a nexus of the global supply chain in maritime trade, and to guarantee a secure transit procedure by reducing material waste. The research encompasses theoretical computations for the bunker situated near the port of Samsun; many engineering investigations were conducted, and design evaluations were executed utilizing the finite element approach in response to seismic impacts. FEM (Finite Element Method) analyses determined the ground bearing capacity, incorporating the seismic factor of the bunker’s location. The seismic loads impacting the structure were assessed through displacement and irregularity evaluations at the critical values of static and dynamic load combinations anticipated in the design. The study results were validated in accordance with the general steel design code ÇYTHYEDY 2018 (YDKT) [1] and the Turkish Building Earthquake Code (TBDY 2018) [2]. Improvements to the ecological port loading bunker design were implemented based on conclusions and theoretical validations derived from the results, leading to the establishment of a secure structural component inside the port. The undertaken work would enhance the 'Green Port Project for Turkish Ports' established by the Ministry of Transport and Infrastructure of the Republic of Turkey, as it encompasses measures designed to mitigate the adverse environmental effects of ports. The system devised in the study is wholly domestically manufactured and will diminish environmental pollution from dusting by a minimum of 85%, establish eco-friendly transmission zones in ports, and, due to its recovery capability, reclaim approximately 75% of material losses during loading. It exemplifies a solution in addressing environmental and climatic issues within the context of the European Green Deal, which delineates the European Union's objectives for implementing sustainable economic strategies globally.

The Effects of Trisiloxane and Polyhydroxycarboxylic Acids (PHCA) on Reducing the Water Footprint in Cotton (Gossypium hirsutum L.) Cultivation

2026-01-12T13:17:33+03:00

Cotton is one of the most important industrial crops produced worldwide. However, in our country as well as in many others, cotton cultivation requires a substantial amount of water. Nowadays, in addition to improving cotton quality, reducing the water footprint during cultivation has become a major priority. This study focused on reducing the water footprint in cotton cultivation by utilizing Trisiloxane and Polyhydroxycarboxylic Acids (PHCAs). Two main experimental groups were established: the first group received 100% irrigation, and the second group received 75% irrigation.Each group included four sub-treatments: 0 L/da (control), 0.5 L/da, 1 L/da, and 2 L/da of Trisiloxane + PHCA. The experiment was conducted in controlled climate chambers maintained at 25 °C, with constant light and humidity conditions.At the end of the experiment, growth parameters of the potted cotton plants were measured. Trisiloxane facilitated water spreading in the soil, while PHCAs enhanced nutrient uptake by the roots. As a result, the combined use of Trisiloxane and PHCAs reduced the water footprint by 25%, demonstrating significant efficiency in optimizing water usage in cotton cultivation.

Tire Cavity Noise Reduction by Using Helmholtz-Based Sandwich Resonator

2026-01-12T11:49:14+03:00

With the elimination of internal combustion engines in electric vehicles, noticeable changes have occurred in the in-cabin noise profile. The absence of engine noise has made road and tire-induced noises more prominent, leading to the emergence of passive acoustic issues such as cavity noise. Cavity noise is a distinct type of noise that negatively affects interior comfort, caused by the resonance of the enclosed air volume between the wheel and the tire.
In this study, tire-induced noise types are first classified in general terms, and then the physical basis of cavity noise is explained through the Helmholtz resonator model. Existing solutions in the literature are examined, and as an alternative, a modular and highly volume-efficient sandwich resonator design that can be integrated into all wheel types is proposed. This design differs from similar studies by maximizing the utilization of the gap volume between the wheel and the tire and being easily adaptable to different wheel geometries.

Single-Bath Dyeing of Blends of Cotton Fibers with New Generation Polyacrylonitrile Fibers with Reactive Dye in Line with the Target of Sustainable Production

2025-12-04T16:10:30+03:00

As is well known, today’s industrial dyeing of cotton/polyacrylonitrile blend fabrics is done first with cationic dyes on the acrylic side, followed by reactive dyes on the cotton side in the second bath. This two-bath, two-step process consumes high amounts of water, energy, and chemicals. The only way to dye acrylic/cotton blends with a single dye is to make the fibers dyeable with the same dye. To make these blends dyeable in a single bath, modified polyacrylonitrile fibers that can be dyed with cotton dyes have been introduced. This study investigated the single-bath sustainable dyeability of acrylic/cotton blend fabrics produced using Oncedye Acrylic™, an acrylic fiber developed by Aksa Acrylic that can be dyed with reactive dyes. The findings indicate that the dye type, dyeing depth, and pretreatment are critical for achieving a truly uniform color. It can be concluded that, in particular, when only scouring is used without bleaching, the acrylic and cotton are dyed more closely, reducing the problem of graininess. It is also worth noting that a more

Open-Source LLM Integrated Data Analysis Assistant with Tableau

2026-01-12T11:49:32+03:00

In corporate data analysis processes, the ability for users to perform data querying and data analysis using natural language, without needing technical knowledge, has become a critical requirement, especially for mid and senior-level managers. This study proposes a solution that offers natural language interaction in TR/EN languages and meets internal data security requirements. The platform connects to Tableau data sources through open-source LLM integration and communicates with data sources published in Tableau via the VDS API to provide real-time analysis and predictions. The architecture also has the flexibility to be integrated with cloud-native AI services in the future. The solution, with its self-service ease of use, enables data analysts and decision-makers to obtain rapid insights without having to worry about technical details.