KJCKorea
Journal Central

This study proposes an HAST architecture that integrates a machine learning–based LSTM model with SAST and DAST to address the growing number of vulnerabilities in web application environments. An analysis of previous studies reveals several limitations in existing web vulnerability detection approaches, including the lack of standardized datasets, limited domain generalization, and insufficient responsiveness to real-time attack scenarios. To overcome these challenges, the proposed architecture combines LSTM-based request sequence analysis with a unified SAST–DAST pipeline. The proposed HAST structure supports real-time request detection, coordinated static and dynamic analysis, and a retrainable expansion mechanism, enabling a stepwise response to evolving web application environments and emerging attack patterns. The results are expected to support the development of an integrated response framework for web vulnerability detection and to provide a structural design foundation for future research.

While Single Image Super-Resolution (SISR) using paired High-Resolution (HR) and Low-Resolution (LR) datasets is effective, obtaining these pairs in real-world scenarios is a challenge. We address this by applying SR to 'Data Counting for OLED Degradation Compensation,' a domain where acquiring such real-world datasets is feasible. This approach uses SR to efficiently reduce the large amount of data required for the counting-based compensation method, which tracks image accumulation to counter OLED deterioration. For our task, we selected the most suitable model and employed a Sequential Up-scaling technique to effectively implement high-magnification SR.

As AI recommendation services become widespread, concerns about fairness, trust, and transparency have become central to understanding user acceptance. This study analyzes how users’ fairness perception influences trust and continuance intention, and assesses the moderating and mediated roles of algorithmic transparency using data from the 2024 Intelligent Information Society Panel Survey. PLS-SEM results indicate that fairness perception significantly enhances both trust and continuance intention, and that trust strongly predicts continued use. Trust also mediates the fairness–continuance link, while algorithmic transparency strengthens the fairness–trust relationship. These findings highlight the importance of the fairness–trust–transparency interplay in shaping user responses and provide practical implications for transparent algorithm design and improved accessibility for vulnerable groups.