Tech-E

User Experience Analysis of the Goods and Services Procurement Agency Website Using the Heuristic Evaluation Method

Rasmila — 2026-02-02

This study evaluates user experience (UX) on the South Sumatra Province Goods and Services Procurement Bureau (PBJ) website using Jakob Nielsen’s Heuristic Evaluation to identify usability barriers affecting e-government transparency and accountability. As public procurement increasingly relies on digital platforms, preliminary assessments revealed critical issues including unintuitive navigation, limited system feedback, and inadequate user guidance that risk undermining public trust. The evaluation employed a 30-item Likert-scale questionnaire aligned with ten heuristic principles and was conducted by expert evaluators. The results show an overall usability index of 81, classified as excellent, with strong performance in consistency and standards (92 percent) and aesthetic and minimalist design (88 percent). Nevertheless, lower scores emerged in help and documentation (69 percent) and error prevention (78 percent), indicating priority areas for improvement. The findings confirm that usability heuristics function as governance enablers rather than purely technical attributes. Improved visibility of system status and effective error recovery mechanisms enhance procurement oversight and are associated with a 20 to 30 percent reduction in vendor drop-off rates. Contextual tooltips and guidance further strengthen accountability in high-value public tenders. The study recommends simplifying navigation structures, strengthening form validation, and implementing searchable frequently asked questions for sustainable provincial e-government implementation initiatives.

Optimization of CNN and Vision Transformer Models in Addressing Long-Tailed Data Imbalance for Satellite Cloud Image Classification

Revatta Manggala Nandivadhano — 2026-02-02

This study investigates long-tailed satellite cloud image classification by comparing CNN and Vision Transformers (ViT) built upon vision–language foundation models. A large-scale satellite cloud dataset with 11 highly imbalanced classes, including a dominant non-phenomenon category, is used to represent realistic atmospheric variability. The data are split using stratified sampling, standardized to a fixed resolution, and used to fine-tune CLIP-based backbones from RemoteCLIP and GeoRSCLIP through parameter-efficient adaptation. Several loss functions Cross Entropy, Logit Adjustment, Focal, Class-Balanced, and label-distribution–aware variants are evaluated, along with experiments examining majority-class removal and adapter bottleneck adjustments. Initial results show that Logit Adjustment causes majority-class collapse under default settings. After optimization, ViT-based models consistently outperform CNN models, achieving higher accuracy and more balanced macro-level performance. Class-Balanced loss emerges as the most effective objective, offering a strong trade-off between overall accuracy and per-class fairness. Increasing the adapter bottleneck dimension further boosts ViT performance, enabling the best configuration to match or exceed prior benchmarks while improving minority-class recognition. The final optimized model is deployed in a web-based prediction system, demonstrating the practical potential of foundation-model approaches for satellite-driven weather analysis.

Design and Implementation of an Integrated Internal Collaboration and Communication Platform Using the SDLC Waterfall Method

Riyan Permana — 2026-02-02

Collaboration and communication platforms play a critical role in supporting modern organizations, particularly those operating in hybrid and remote working environments. These platforms enhance workflow efficiency, improve coordination across teams, and facilitate seamless information exchange. This study aims to design and implement an integrated internal collaboration and communication system that unifies chat, discussion forums, file sharing, and internal video conferencing within a single platform. The system was developed using the Software Development Life Cycle (SDLC) with the Waterfall model, which includes requirement analysis, system design, implementation, and testing phases. Data collection was conducted through direct observation and structured interviews to identify user needs and determine system requirements. Functional testing was carried out to assess system performance, interface responsiveness, and feature reliability. The results indicate that all major functionalities work effectively, including real-time messaging, collaborative document editing, centralized file management, and virtual meeting capabilities. Overall, the proposed platform enhances coordination efficiency, reduces data fragmentation, and minimizes reliance on external third-party applications. This research provides a comprehensive and structured blueprint for developing integrated internal collaboration systems that support effective communication and centralized information management within organizational environments.

Simulation Approaches To Thermal Comfort In Tvet Workshops: A Review

Norarziana Sasudian — 2026-02-02

Thermal comfort is a crucial element of indoor environmental quality (IEQ), affecting health, cognitive performance, and educational results. Technical and Vocational Education and Training (TVET) workshops exhibit unique thermal issues owing to heat-producing equipment, elevated occupant activity, and intricate ventilation systems. This study consolidates simulation-based studies published from 2016 to 2025 regarding thermal comfort in educational and vocational settings. This narrative review examines the uses of Computational Fluid Dynamics (CFD), EnergyPlus, DesignBuilder, and TRNSYS. The findings indicate that while simulation tools are extensively utilized in classrooms and laboratories, their application in TVET workshops is still constrained. Some of the most important shortcomings are that equipment heat loads are too accessible, occupant activity is not extensively represented, field validation is not effective enough, and hot and humid climates are not given enough attention. The paper supports representative TVET workshop models, hybrid simulation–measurement workflows, and Education 5.0-aligned simulation technologies in vocational education.

FireReady Challenge: A 2D Gamified Prototype for Community Fire Safety Preparedness

Rabeah Md Zin — 2026-02-02

Fire protection remains a critical public concern, as fire-related incidents continue to cause substantial property damage, personal injuries, and loss of life. Although public awareness campaigns and routine safety drills are widely implemented, conventional didactic training methods such as posters, presentations, and demonstrations often fail to generate lasting comprehension or meaningful behavioural change. To address this gap, this paper presents the FireReady Challenge, a 2D gamified learning prototype designed to enhance community awareness and preparedness in fire safety. The game adopts the ADDIE instructional model to systematically translate key learning objectives into interactive digital experiences focused on hazard identification, safe evacuation procedures, emergency communication readiness, and proper fire extinguisher use. Gamification elements, including points, badges, real-time feedback, and time-bound missions, are embedded to strengthen motivation and user engagement. This concept design demonstrates an innovative approach to ICT-based fire safety education for community dissemination. Future research will involve pilot-scale deployment and empirical validation to evaluate learning outcomes, usability, and user acceptance.

Desain Microcomputer Cloud Computing for Informatics Study Program at LIA University Based on Local Area Network (LAN)

Ariya Pannadhitthana Candra — 2026-02-02

Digital transformation in higher education increasingly requires information technology infrastructure that is efficient, flexible, secure, and sustainable. Cloud computing has emerged as a strategic solution by enabling scalable resources, centralized management, and service standardization effectively. However, public cloud adoption in academic institutions frequently encounters constraints, including recurring operational costs, data security and sovereignty risks, regulatory concerns, and strong dependence on reliable internet connectivity. These issues are particularly salient for instructional laboratories that demand continuous access, predictable performance, and institutional control. This study aims to design and implement a Local Area Network (LAN)-based cloud computing system using a microcomputer as the primary server within the Informatics Study Program at LIA University. The research employs a research and development (R&D) methodology comprising needs analysis, system architecture design, implementation, and functional as well as performance testing. An open-source virtualization platform is deployed to deliver Infrastructure as a Service (IaaS) and Software as a Service (SaaS) in a private cloud environment. The results demonstrate that the proposed LAN-based local cloud provides centralized computing and storage services with low latency, high stability, and efficient utilization.

Machine Learning Approaches to Workplace Mental Health: Predicting Treatment-Seeking Behavior Using the OSMI Dataset

Nor Aishah Othman — 2026-02-02

Employee mental health is increasingly recognized as essential for sustainable organizational performance, particularly in technology sectors where work intensity and psychological strain are prevalent. This study leverages machine learning to identify predictors of treatment-seeking behavior using the Open Sourcing Mental Illness (OSMI) dataset, which includes 1,387 anonymized responses from the 2014 OSMI survey. The survey examines employees’ experiences and perceptions of mental health in the global tech industry. Through data cleaning and encoding, key factors influencing help-seeking behavior were identified, including family history of mental illness and work interference due to psychological distress. Two machine learning models, Decision Tree and K-Nearest Neighbour (KNN), were employed for prediction. The Decision Tree model achieved an accuracy of 73%, while KNN attained 100%, suggesting high predictive power, albeit with potential overfitting risks. These findings align with recent studies promoting the integration of AI-driven analytics in workplace wellness programs to detect hidden behavioral trends and enable early interventions. The results demonstrate that machine learning models can offer valuable insights into employee well-being and preventative strategies. Future research should focus on incorporating larger, more diverse datasets and adopting explainable AI (XAI) techniques to enhance interpretability, fairness, and trust in predictive systems for mental health in the workplace.

Public Sentiment Analysis of Free Nutritious Meal Program Discourse on Social Media X Using Support Vector Machine N-Gram Features Based

Daniel Silaban — 2026-02-02

The Free Nutritious Meal Program is a government policy aimed at improving the nutritional quality of society and has generated diverse public responses on social media. This study aims to analyze public sentiment toward the Free Nutritious Meal Program on social media X using the Support Vector Machine (SVM) algorithm with N-Gram features and Term Frequency–Inverse Document Frequency (TF-IDF) weighting. The data were collected through a crawling process from social media X, resulting in 1,014 tweets. After data cleaning, 931 tweets were obtained and labeled into two sentiment classes, namely positive and negative. The research stages include text preprocessing, N-Gram feature extraction (unigram and bigram), classification using the SVM algorithm, and model evaluation using the 10-Fold Cross-Validation method with the assistance of the RapidMiner tool. The experimental results show that the SVM model achieved an accuracy of 79.59%. Although the precision value for the negative class is relatively high, the recall and F1-score remain relatively low due to the imbalance in data distribution. Overall, the results indicate that public sentiment toward the Free Nutritious Meal Program on social media X is dominated by positive sentiment. The findings of this study are expected to serve as an initial evaluation for the government in understanding public perceptions of the implementation of the program.

Architectural Analysis of the Repository Pattern in Web-Based Credit Score Conversion Assessment System Based on PermenPAN-RB No. 1 of 2023

Nouval Trezandy Lapatta — 2026-02-02

PermenPAN-RB Regulation No. 1 of 2023 introduced a major shift in functional position assessment by emphasizing performance predicate conversion in credit score evaluation, which increases architectural demands on supporting information systems. In practice, many assessment systems remain tightly coupled and difficult to evolve when regulatory rules, integration sources, or reporting formats change. This paper presents an architecture-oriented analysis of a web-based credit score conversion assessment information system that applies the Repository Pattern as a core architectural mechanism to decouple business logic from persistence, integration, and document-generation concerns. The analysis adopts a scenario-based evaluation approach inspired by the Architecture Tradeoff Analysis Method (ATAM) and is grounded in the ISO/IEC 25010 software quality model, focusing on maintainability, modifiability, testability, scalability, and reliability. Architectural evaluation is conducted by examining layered boundaries, repository abstractions, and dependency injection mechanisms under representative regulatory-driven change scenarios, including rule adjustments, data integration extensions, and reporting modifications. The results demonstrate consistent change localization across architectural layers, where rule changes are confined to service modules, integration changes are absorbed by repository adapters, and reporting changes remain isolated within document-generation components. These findings show that repository-based architectures significantly reduce coupling, improve change isolation, and support the sustainable evolution of government information systems operating under dynamic regulatory environments.