Earticle

This paper analyzes the effect of multicast communication on reducing data traffic in a hybrid MMOG(Massively Multiplayer Online Game) architecture that combines the technical advantages of both client-server and P2P(peer-to-peer) models. In this hybrid architecture, the entire game is divided into multiple regions of equal size, and the server designates a specific player in each region as the region server. Even when the number of concurrently active players increases rapidly, the system aims to prevent bottlenecks on the central server by appropriately utilizing both the server's bandwidth and the bandwidth of the players participating in the P2P network, thereby enabling a cost-effective and scalable MMOG with consistent service quality. In this paper, multicast communication is applied in two cases: when the central server transmits the game state to multiple region servers, and when a specific region server retransmits this game state to players within its region. The paper compares and analyzes the bandwidth usage of multicast communication and unicast communication in these scenarios through simulation. In this hybrid MMOG architecture, the intention is to minimize the use of the central server's bandwidth and processing power by leveraging more of the players' bandwidth and processing capabilities.

In coffee cultivation, accurately classifying fruit ripeness to ensure harvest quality is an important task. In particular, traditional manual harvesting methods in smart farm environments are labor-intensive and prone to errors. In this study, we propose an improved YOLOv11-based system for automatically detecting coffee fruit ripeness in hydroponic environments. The proposed approach consists of two parts: class imbalance mitigation through Instance-Aware Repeat Factor Sampling (IRFS) and network architecture optimization. IRFS improves the sampling weight of rare classes by simultaneously considering class distributions at both the image level and instance level, overcoming the limitations of existing Repeat Factor Sampling, which only considers image-level frequencies. The network architecture uses EfficientNetV2 with MBConvolution blocks as the backbone for improved feature extraction. The neck combines the C2PSA block, which integrates the Cross Stage Partial with Partial Self-Attention mechanism, and the head combines the Dynamic Adaptation Head for adaptive detection. An RGB image dataset was collected from a smart farm in Siheung, Gyeonggi Province, South Korea, and consists of four maturity classes: unripe, semi_ripe, ripe, and overripe. The proposed system achieves 95% on the mAP@0.5 metric, demonstrating superior performance compared to existing YOLO models. Compared to existing models without IRFS, it shows a 5% improvement in performance compared to YOLOv11n. Therefore, consistent performance improvements are observed when applying IRFS to all YOLO models. These results indicate that the combination of class imbalance mitigation and an optimized network architecture is effective for accurate coffee berry maturity detection.

This study investigates how consumer innovativeness and resistance to innovation jointly shape mobile keyword search behavior during pre-purchase decision-making. Drawing on innovation resistance theory and spreading activation theory, we analyze mobile log data from 128 smart-device purchasers using semantic network analysis and sentiment-based keyword classification. Four consumer groups were segmented based on innovativeness and resistance levels. Results show that highly innovative but resistant consumers exhibit fragmented and risk-centered search networks, while those with low resistance form interconnected, exploratory structures. Functional inconvenience prompts pragmatic searches, whereas anxiety leads to emotionally charged expressions. These findings offer theoretical insights into dual psychological dispositions in innovation contexts and practical implications for adaptive digital communication.

Sustainable luxury may evoke a sense of perceived aesthetic unfitness among traditional luxury shoppers; however, the potential impact of this perceived aesthetic unfitness—arising from the integration of sustainability into luxury—on consumption has been rarely examined. The present study explores the role of perceived aesthetic unfitness of sustainable luxury fashion in relation to subjective perspectives of luxury consumption (positive/negative) and brand attitude. Data was analyzed through structural equation modeling. The results reveal that, as predicted, both the negative and positive motivations for luxury significantly increase sustainable luxury perceived aesthetic unfitness. Perceived aesthetic unfitness leads to a significant increase in brand attitude. The results of this study provide essential information for luxury brand managers who want to incorporate sustainability into their product lines. The study offers detailed managerial implications by connecting its results to the existing body of research on luxury and sustainable consumption. The result that both positive (e.g., hedonic, self-expressive) and negative (e.g., conspicuous, materialistic) luxury motivations increase perceived aesthetic unfitness of sustainable luxury products aligns with earlier observations that luxury consumers maintain high aesthetic standards and may be resistant to design deviations from traditional luxury codes. While motivations differ in nature, both types appear to amplify sensitivity to aesthetic incongruities, particularly when sustainability is perceived to compromise visual refinement.

This study proposes an intelligent prediction framework that links the Digital Twin (DT) to an edge Battery Management System (BMS) structure based on Federated Learning (FL) to enhance the fault prediction and real-time response capabilities of battery-based Internet of Things (IoT) systems. The proposed system locally learns a lightweight anomaly detection model using battery status data collected from an edge gateway and comprehensively updates the global model through federated learning. Additionally, the reliability of the prediction results is enhanced by verifying various fault scenarios through digital twin simulation. The experimental results demonstrate that the proposed technique outperforms both the existing centralized method and the general federated learning method in terms of prediction accuracy and alarm response time. This demonstrates that the combination of a field-based, distributed learning structure and a real-time verification system is effective in enhancing battery safety.

Spotify operates under a freemium model that relies heavily on converting users to Premium subscribers for sustained revenue growth and market competitiveness. However, the music streaming platform faces increasing challenges in its conversion funnel. This paper evaluates Spotify’s marketing strategies using the AIDA (Attention, Interest, Desire, Action) framework, focusing on website design, Search Engine Optimization (SEO), social media campaigns, and reputation management. Our analysis shows that Spotify’s website design is cluttered, with multiple call-to-action buttons diluting user focus, unlike competitors with simpler designs. Its SEO and SEM efforts drive awareness but yield unclear conversion results. Social media campaigns show high engagement but lack consistency and platform-specific features. Reputation management is effective but hindered by slow responses and negative reviews. To encourage user transition to Spotify’s Premium service and reduce churn, the company should simplify its website, refine pricing in emerging markets, enhance onboarding to highlight Premium benefits, and improve social media and reputation management strategies. These improvements can strengthen user retention and Spotify’s competitive position.

As more young adults postpone or forgo marriage, challenges in early-stage romantic communication— particularly on first dates—are becoming more evident. This study introduces a conversation simulation that uses AI-generated user personas to help individuals feel more at ease during such encounters. To create realistic personas, the researchers designed a survey referencing data formats from Korean matchmaking services. The responses were processed using ChatGPT to model users with distinct personality traits. These personas were then used in first-date simulations built with GPT-4.o and Python. The simulated conversations followed prompts that incorporated three elements: an overall context, conversation rules, and discussion topics. The simulation was generally effective, improving the natural flow of conversation and topic engagement. The most successful exchanges occurred between users with shared interests or similar temperaments. This AI-based model holds promise as a rehearsal tool for those preparing for real-world dating or marriage scenarios. Future development could enhance realism and evaluation by including nonverbal cues, such as facial expressions and tone of voice, and by expanding language support.

Mobile digital platforms enable rapid large-scale sales of hotel products in the Chinese market. A case in point is Alibaba’s hospitality platform Fliggy and The Universal Studios Grand Hotel that opened in Beijing in 2021. Hotel rooms sold so rapidly that the platform crashed. Fliggy sold to 100,000 customers within half an hour. A first purpose of this study was to explain how digital platforms in China enable such sales achievement. How are luxury hotels able to strategically achieve rapid mass sales in the Chinese market? A second purpose was to investigate how hotels compete in the Chinese digital platform market in comparison to other international platform markets, and with regard to their platform business models and sales strategies. To achieve this objective, data was collected through 17 semi-structured interviews with hotel executives, platform owners, and destination marketing organizations, and analyzed using a qualitative content analysis. The results contribute to the understanding of the Chinese hotel platform market and sales strategies of luxury hotels in China. Platforms in China enable hotels in the high-scale luxury market to achieve mass sales through product differentiation. Digital platforms reduce transaction and search costs, and the marketing costs of brand exposure and sales channels. This contrasts to traditional hotel business models where mass selling is associated with lower costs from the mass production of functional standardized products. Mass selling has accordingly traditionally been associated with lower-scale hotels. Mobile platforms enable the rapid mass selling of luxury hospitality products through product differentiation and by reducing search costs in the large Chinese customer market.

This paper intends to analyze the current status of success factors in the domestic franchise business using big data. Franchise businesses that occupy most markets can continue to operate stably using minimal capital, but competition is occurring due to overflowing franchise businesses and the exit rate is also increasing. In this dual situation, the franchise business needs empirical analysis of successful operation strategies, and we intend to derive success factors by utilizing big data for future development. Using TexTom, a big data analysis program, data were collected from May 2022 to May 2025 and text mining was carried out, and based on this, analysis results were derived more easily using the connectivity analysis tool UCinet visualization function. As a result of the study, key factors such as "improvement of operating system", "strategic support of member headquarters", "strengthening brand image", "sensitive response to consumer trends", and "strategies for overseas expansion" were derived. Such results will greatly contribute to proposing strategic directions to franchise management practitioners in the future.

The era of digital transformation is reshaping society, and the resulting digital divide among generations has become a critical issue. Seniors, in particular, experience difficulties in digital literacy and access, which affect daily life, social participation, and use of welfare services. We designed this study to analyze public perceptions of senior digital transformation using social big data, with the aim of deriving policy, educational, and technological implications. A total of 3,841 online documents were collected from Naver, Daum, and Google News between May 2023 and May 2025. Text mining was conducted using the keyword “Senior + Digital Transformation,” and the top 50 keywords were extracted using Textom. Semantic network analysis and CONCOR clustering were performed using UCINET 6 and NetDraw. The results revealed four major discourse clusters: policy and acceptance, service and industry environment, regional disparities, and digital competency education. These findings suggest that the issue is not merely technological but structurally social. This study offers distinctive insights by structurally analyzing social discourse, and recommends follow-up research such as qualitative interviews and policy evaluations.

In the digital age, TikTok’s short-form videos have reshaped customer engagement, particularly in the tourism industry. This study examines the impact of compelling content factors on customers’ emotional engagement, and how this emotional engagement, in turn, drives behavioral engagement. Data were collected from 261 TikTok users in Vietnam through an online survey. The results indicate that creative content, highquality content, and trending content significantly influence emotional engagement, which subsequently affects behavioral engagement. However, content enjoyment and attractiveness were not found to have a significant impact on emotional engagement. This study provides both theoretical and practical insights for businesses aiming to enhance their social media marketing strategies effectively.

This study originates from the need to analyze how kimchi, a traditional Korean fermented food gaining global attention amid the Korean Wave and health trends, is being utilized as a global expansion strategy in the food service industry. Kimchi is being re-evaluated as a strategic asset encompassing health, culture, and sustainability, and this research aims to identify regional characteristics and trends of kimchi-based marketing strategies by focusing on the interests and responses of consumers in the digital age. The study analyzed various online texts collected from social media, reviews, and news sources worldwide between January 2022 and December 2024 using the Textom platform, visualizing frequently mentioned words and examining the relationships between co-occurring terms to understand consumer interest patterns. Unlike traditional methods such as surveys or interviews, this study attempted a new approach by analyzing real-time responses based on voluntarily posted online texts, showing that keywords such as ‘kimchi’, ‘globalization’, ‘localization’, ‘fusion cuisine’, and ‘global marketing’ were central, reflecting value-centered consumption trends. Future research should enhance methods for more sophisticated analysis of such digital text data and expand the scope of research to other Korean foods beyond kimchi to deepen strategic insights for the globalization of Korean cuisine.

In the current animation production environment, data compatibility between different tools has become increasingly important. To effectively reuse character motion data, an accurate retargeting method between different skeletal structures is essential. This paper explores how motion data created in various contexts can be accurately retargeted to MetaHuman characters in Unreal Engine. Utilizing Maya's HumanIK system, we aimed to develop a Python-based automation script that converts character rigs with motion data created in various tools into HumanIK rigs, enabling precise retargeting to MetaHuman characters. The proposed system provides a retargeting pipeline and scripts designed to resolve common issues such as positional offsets, arm rotation errors, and foot sliding. Ultimately, through actual implementation, the system demonstrated improved efficiency in animation production.

With the rapid development of generative artificial intelligence technologies, traditional workflows in 2D image creation and 3D modeling are being restructured through intelligent approaches. This paper proposes an automated modeling pipeline based on the DeepSeek text generation system and AI-based image/model transformation tools. The pipeline is designed to initiate from textual descriptions, generate corresponding 2D images through AI illustration, and then automatically convert them into 3D models with texture mapping. Empirical evaluations demonstrate that this workflow not only significantly enhances modeling efficiency but also lowers the entry barrier to 3D modeling, offering a novel path for content generation in virtual reality, digital art, and the gaming industry.

This study explores the development and implementation of a learner-centered instructional model that integrates Project-based Learning (PjBL) with Design Thinking methodologies in the context of a university-level Visual Communication Design course. The instructional innovation was applied to the course “Design” during the second semester of the 2024 academic year. The study aimed to improve students’ creative problem-solving skills, collaboration, and self-directed learning through a structured project cycle grounded in real-world challenges. The results demonstrate that the integration of PjBL and Design Thinking significantly enhanced student engagement, critical thinking, and the ability to generate practical design outcomes. Qualitative feedback and output evaluations further support the pedagogical effectiveness of the model.

This study proposes a novel model structure for implementing 3D lip-sync by extending the Korean speech synthesis model, Korean-FastSpeech2. Existing Korean lip-sync technologies have struggled to accurately render the three-dimensional expressions of the Korean pronunciations "아" (/a/), "오" (/o/), and "우" (/u/), particularly the lip rounding (mouthPucker). To address this, we introduce a Lip Predictor to the Encoder-Variance Adaptor-Decoder architecture, enabling the model to learn ARKit data. The Lip Predictor, built on a Transformer decoder with four layers and eight multi-head attentions, processes phoneme features and temporal information. By sharing the Variance Adaptor’s output with the speech output Decoder, it naturally resolves synchronization issues between speech and lip movements, which is the core contribution of this study. The proposed model facilitates specialized learning for "아", "오", and "우" pronunciations and is expected to offer superior precision, synchronization accuracy, and scalability compared to existing 3D lip-sync algorithms such as Audio2Face, VOCA, and FaceFormer. This work highlights the potential for advancing lip-sync technology for minority languages.

Effective fusion of heterogeneous modalities is a critical factor for improving performance in multimodal learning. However, existing Cross-Modal Transformer (CMT)-based fusion methods are constrained by fixed attention paths, which reduce adaptability to diverse inputs and restrict flexible exploration of optimal fusion strategies. Furthermore, as the number of modalities increases, the computational complexity grows exponentially, leading to scalability bottlenecks. To address these limitations, we propose a Multimodal Routing Optimization (MRO) framework that restructures cross-modal attention paths as a Supernet structure, drawing inspiration from the Once-for-All (OFA) paradigm. Leveraging Neural Architecture Search (NAS), Multimodal Routing Optimization (MRO) dynamically selects optimal routing paths that balance accuracy and computational cost (FLOPs), enabling scalable and efficient multimodal fusion.

This study aimed to clarify the effects of parental intimacy and optimism on middle school students’ learning motivation and school life satisfaction through the mediating variable of learning motivation. The relationship between children and their parents is an important variable that affects school life. It also affects personality formation. To this end, data collected through a questionnaire survey of 380 middle school students in Gyeonggi Province, who were convenience-sampled, were analyzed. The research questions set for this study were to examine the differences in school life satisfaction according to the general characteristics of the subjects and to clarify the relationship among parental intimacy, optimism, learning motivation, and school life satisfaction. The results of the study are as follows. As a result of analyzing the differences in middle school students' satisfaction with school life, it was found that there were significant differences in grade and parental education level. The first graders showed higher satisfaction with school life than the second and third graders, and the higher the parental education level, the higher the satisfaction with school life. In addition, as a result of examining the significance of the estimated coefficient for the research model, it was found that parental intimacy and optimism affected middle school students' satisfaction with school life through learning motivation. Specifically, parental intimacy and optimism increased middle school students' motivation to learn, and the higher the motivation to learn, the higher the satisfaction with school life. Therefore, in order to increase middle school students' satisfaction with school life, it is necessary to increase parents' intimacy with their children and for middle school students themselves to make efforts to be optimistic.

We have experimentally confirmed that image learning model environments are vulnerable to backdoor attacks that cause misclassification through trigger insertion. Backdoor attacks were influenced by the tradeoff between detection evasion and image quality maintenance, as well as attack stability and disturbance characteristics depending on the multi-resolution environment. We designed and experimented with a total of 10 attack techniques—InputAware, Reflection, LIRA, NeuralCleanse, AdaptiveTrigger, and their hybrid attack models—targeted at images with resolutions of 224× 224, 512× 512, and 1024× 1024. In this research, we used performance metrics such as attack success rate (ASR), PSNR, SSIM, and confidence, and compared and analyzed performance differences according to resolution changes. The experimental results showed that AdaptiveTrigger and Hybrid AdaptiveTrigger achieved 100% attack success rate at all resolutions and demonstrated high attack risk. In particular, the Hybrid InputAware model demonstrated the most balanced performance, showing a balance between success rate and stealthiness, as well as strong stability even with resolution changes. Through this study, we have comprehensively analyzed the threat level and evasion capabilities of various backdoor attack techniques, as well as changes in attack characteristics due to resolution changes. We expect that this research will contribute to the design and defense of attack detection systems targeting image learning in the future.

In this paper, a folded dipole antenna operating in the ISM band is presented. The antenna was designed by adjusting the ratio of the folded dipole arms and incorporating a matching stub technique. Each element of the antenna was optimized through a parameter sweep. The analysis confirmed that the slot length in the folded dipole and the width of the matching stub enable fine-tuning of the operating frequency. Furthermore, the antenna characteristics can be effectively optimized by adjusting the position of the matching stub and the arm ratio of the folded dipole. The antenna has a compact size of 60 mm × 38 mm × 3 mm and is fabricated on an FR4 substrate. It achieves a 10 dB bandwidth from 2.37 to 2.51 GHz, corresponding to a bandwidth of 140 MHz. The radiation pattern resembles that of a half-wavelength dipole, and the maximum gain is 1.59 dBi. Based on the results of this study, the proposed antenna is considered suitable for applications in IoT, ZigBee, and smart factory environments.

AI platforms have been widely introduced in East Asian rural revitalization to enhance agricultural efficiency and promote smart governance. However, many of these platforms encounter "build-but-not-use" failures at the grassroots level. Drawing on comparative insights from Korea and Japan, as well as 13 semi-structured interviews conducted in Fujian, China, this study identifies three core institutional obstacles: information disconnection, feedback deficiency, and responsibility mismatch.To systematically analyze these challenges, a four-level nested framework— encompassing macro, meso, micro, and technical levels— is developed, alongside three mechanism chains: information, feedback, and responsibility. By integrating policy analysis, digital governance theory, and qualitative field research, the study constructs a structural diagnostic framework to reveal misalignments between institutional configurations and platform performance.Findings demonstrate that technical inefficiencies are deeply rooted in institutional coordination gaps rather than technological limitations alone. The research further validates three disconnection mechanisms— information gaps, feedback disruption, and responsibility ambiguity— across diverse East Asian contexts. This transdisciplinary approach bridges public administration and information systems, offering a novel pathway for institutional adaptation. The study offers transferable insights for optimizing rural digital governance and highlights the importance of cross-level coordination in the sustainable implementation of AI platforms.

This study investigates the structural characteristics and temporal evolution of the Pangyo AI Cluster’s interfirm networks using Social Network Analysis (SNA). As traditional cluster policies are rooted in physical proximity and manufacturing-oriented supply chains, their applicability to the AI industry—driven by intangible assets and decentralized collaborations—remains uncertain. Using transaction data of 528 AI firms (198,327 transactions from 2021–2023), the study analyzes centrality indicators to compare the Pangyo AI Cluster with conventional manufacturing clusters. Results reveal that while manufacturing clusters exhibit a vertically integrated, conglomerate-driven network, the Pangyo cluster displays a more horizontal, diversified structure involving SMEs and public institutions. Furthermore, the study finds that public agencies, such as the Seongnam Industry Promotion Agency, play a critical intermediary role, underscoring the importance of policy support in early-stage ecosystems. These findings suggest the need for differentiated, digitally focused cluster policies tailored to AI industries, moving beyond physical co-location models. This research provides theoretical and empirical insights into designing future-ready innovation clusters.

This study empirically evaluates the business feasibility of a Software-as-a-Service (SaaS) model in the context of the industry's transition toward Software-Defined Vehicles (SDV). Focusing on the “Intelligent Wiper Service” — a subscription-based software service optimized for SDV environments — the research employs a Pro Forma analysis to assess profitability and payback periods from both single customer and multi-customer perspectives. The results demonstrate a high Customer Lifetime Value (LTV) relative to Customer Acquisition Cost (CAC), and simulation outcomes across varying churn rates and subscription pricing scenarios suggest strong potential for global market expansion. Furthermore, this study presents a conceptual model of the Intelligent Wiper Service and highlights the synergy between SDV-based vehicle SaaS and the subscription economy, offering a sustainable revenue framework for automotive OEMs.

This study aims to analyze various types of audio recognition attacks (Noise, Reverse, Amplify, Attenuate, Shift, etc.) faced by AI-based speech recognition systems and their effects. Through experiments, the impact of each attack type on the output of the speech model was quantitatively evaluated using performance metrics such as MSE, MAE, SNR, CrossCorrMax, CosineSim, PearsonCorr, Emotion_Label, and Emotion_Score. The results showed that Reverse and Shift attacks severely degraded the emotion classification and reliability of the speech recognition model, while Amplify and Attenuate attacks caused subtle but significant changes in emotion labels. This study aims to analyze various types of audio recognition attacks (Noise, Reverse, Amplify, Attenuate, Shift, etc.) faced by AI-based speech recognition systems and their effects. Through experiments, the impact of each attack type on the speech model's output was quantitatively evaluated using evaluation metrics such as MSE, MAE, SNR, CrossCorrMax, CosineSim, PearsonCorr, Emotion_Label, and Emotion_Score. The results showed that Reverse and Shift attacks significantly degraded the emotion classification and reliability of the speech recognition model, while Amplify and Attenuate attacks caused subtle but important changes in emotion labels.

We purposed to investigate the changes in internal training load and shot performance during a 9-hole round of golf in KPGA professional golfers, with the goal of elucidating the influence of physiological and psychological responses on performance. Eight KPGA-certified professional golfers participated in a simulated 9-hole round. Heart rate variability (HRV), blood lactate concentration, and shot metrics were measured at the teeing grounds of holes 1, 5, and 9. Driver shot speed, carry distance, and shot accuracy were assessed and analyzed using repeated measures one-way ANOVA. HRV indicators showed a decreasing trend as the round progressed, while blood lactate concentration peaked at hole 5 and slightly declined by hole 9. Shot analysis revealed the highest ball speed and the lowest shot error at hole 5, indicating improved performance at mid-round. Moderate levels of physiological tension and fatigue may positively contribute to golf performance. HRV and blood lactate levels are suggested to be valuable physiological markers for monitoring player condition and informing strategic decision-making during competitive play.

This study proposes an AI based on abnormal behavior detection method to identify abnormal transactions that may occur in blockchain based data transactions. For the study, This study utilized the Hyperledger Fabric platform and automatic transactions of smart contracts. The traded data is recorded on the blockchain through smart contracts, and sensitive metadata is stored in off chain storage. As an AI model, XGBoost, which has high interpretability, was used, and abnormal transactions are detected by analyzing user transaction logs through the model, and the trust score is adjusted in real time through smart contracts. The experimental results showed that the performance indicators were improved by about 2~ 5% compared to previous similar studies, and this study can operate a trust based transaction automation system.

The purpose of this study is to compare and analyze BadNet, Blended, WaNet, FIBA, SIG,CLBA, Label Consistent attack models in terms of their variations in visual recognition, their prediction reliability, and their attack success rate, and to consider the impact of attack models on AI image recognition. In this study, eight representative backdoor attack models were selected, and Confidence, PSNR, SSIM, L1, L2, and L∞ were selected as performance metrics to evaluate the attack performance. The results show that BadNet, Label Consistent, and CLIBA attacks are the most natural attacks in terms of SSIM and PSNR; SIG, CLBA, and Adaptive SIG are attacks that succeed in changing targets with high prediction reliability in terms of Confidence; and FIBA and WaNet are models that are easy to detect with large variations in terms of L1/L2. In this study, we confirmed through experiments that stealth-based attacks such as CLBA, Label Consistent, and Adaptive SIG are attacks that can pose a real threat.

Blockchain is a distributed ledger technology that enables accurate and transparent transactions, _and it is frequently used_ when handling multivariate, high precision data such as medical and financial transactions. When processing data with a blockchain based multivariate structure, dimensionality reduction for unnecessary information and processing efficiency through principal component analysis are required for computational efficiency. In this paper, in order to improve the efficiency of data processing when processing the relevant data on the Hyperledger Fabric platform, PCA (Principal Component Analysis) based covariance structure analysis was applied, and the results were applied as smart contract automatic contract variables to enable efficiency based transactions. Through this paper, _This study_ were able to extract meaningful principal components from multivariate data, maintain trust based data management, and increase efficiency related to computational complexity and data processing speed through lightweight processing, proving that the processing speed was improved by more than 12% compared to the experimental results without PCA application.

This study investigated Chinese adults aged 60 and above, focusing on the relationships among perceived age discrimination, feelings of helplessness, and the exercise adherence, while also examining the moderating role of social support. The findings reveal that perceived age discrimination significantly increases feelings of helplessness, which in turn reduces the exercise adherence—demonstrating a partial mediating effect. Moreover, age discrimination negatively influences exercise adherence both directly and indirectly through helplessness. Importantly, social support serves as a significant moderator. High levels of support effectively buffer the negative effects of age discrimination on both helplessness and exercise adherence, to the extent that the direct and indirect effects become nearly negligible. Based on these results, future research is encouraged to adopt longitudinal designs to better capture dynamic processes, examine the distinct effects of different types of social support, and include more diverse elderly populations to enhance generalizability. Incorporating frameworks such as active aging, self-efficacy, and social belonging could also offer deeper insights into how older adults adapt psychologically and behaviorally in the face of discrimination. Overall, this study enriches the theoretical understanding of the psychological mechanisms shaping exercise behavior in later life and provides valuable guidance for promoting healthy aging through supportive interventions.

This study focuses on the design and performance evaluation of a lightweight object detection model, YOLOv11BiFormer, for the efficient detection of road surface defects such as alligator cracks, longitudinal cracks, potholes, and transversal cracks. To enhance computational efficiency and improve the detection of small-scale objects, the proposed model integrates the BiFormer block and C2f module into the existing YOLOv11 architecture. The dataset used for training and evaluation consists of 7,238 highresolution images, which were evenly divided into 5,065 training images and 2,137 validation images across the four defect categories. Experimental results show that the YOLOv11BiFormer model outperforms the original YOLOv11 in multiple metrics: mAP 0.5 improved from 0.522 to 0.546, mAP 0.5:0.95 increased from 0.691 to 0.703, and precision rose from 0.462 to 0.497. Furthermore, the number of parameters and model size were reduced from 2,582,932 to 2,464,956 and from 5.5MB to 5.2MB, respectively. Visual analysis also demonstrated superior detection accuracy and clearer boundary identification with the BiFormer-enhanced model.These findings suggest that the proposed YOLOv11 BiFormer model is well-suited for real-time road defect detection in mobile devices and edge computing environments, offering a promising solution for intelligent transportation systems and automated infrastructure inspection.