Journal of Software Engineering and Information System (SEIS)

IMPLEMENTASI SISTEM SMART VENTILATION BERBASIS IOT MENGGUNAKAN THRESHOLD KUALITAS UDARA

Mon, 31 Aug 2026 00:00:00 +0700

Poor air quality in enclosed workspaces, particularly in Micro, Small, and Medium Enterprises (MSMEs) that operate CNC laser cutting machines for acrylic processing, poses potential health risks and reduces worker comfort. This study aims to design and implement an Internet of Things (IoT)–based Smart Ventilation system capable of automatically monitoring and controlling indoor air quality based on predefined threshold values. The system integrates air quality sensors to detect smoke or hazardous gases, an ESP32 microcontroller as the central processing unit, and an exhaust fan as the actuator. Sensor data are processed in real time and used as the basis for activating or deactivating the ventilation system, while monitoring information is displayed through an IoT platform for remote supervision. Experimental results indicate that the proposed system can accurately identify changes in air quality and respond automatically to hazardous conditions in accordance with the defined thresholds. The implementation of this system is expected to improve indoor air quality, support occupational health and safety, and enhance energy efficiency through condition-based ventilation control.

ANALISIS PERFORMA SMOTE DAN SMOTE-TOMEKLINK UNTUK KLASIFIKASI DIABETES : XGBOOST-SHAP

Arifah Budi Hidayah, Ali Nur Ikhsan — Mon, 31 Aug 2026 00:00:00 +0700

This study examines the effects of SMOTE and SMOTE-TomekLink on XGBoost classification and SHAP interpretability using an imbalanced diabetes dataset (100,000 samples, 91.5% : 8.5% ratio). Three research gaps are addressed: (i) direct experimental comparison of both resampling techniques within identical pipelines; (ii) quantification of minority sensitivity stability through Stratified 5-Fold Cross-Validation; and (iii) reliability of post-resampling SHAP interpretation. Results demonstrate significant superiority of SMOTE-TomekLink with a F1-score of 0.89 (SMOTE 0.85, baseline 0.81), recall of 0.86 (SMOTE 0.85, baseline 0.70), and the lowest false positive rate of 176 cases (SMOTE 345, baseline 40). The lowest cross-validation standard deviation (0.0021) validates superior stability. These figures meet the very good performance category (F1-score >0.85, recall >0.80) for imbalanced medical classification. Primary contribution: the first controlled comparative design that simultaneously examines resampling effects on classification metrics and SHAP interpretation validity. SHAP analysis verifies HbA1c and blood glucose as dominant predictors (52.3% variance), consistent with medical standards.

PENGEMBANGAN SISTEM PREDIKSI PERSEDIAAN BARANG BERBASIS WEB PADA TOKO MURNI

Muhammad Andika Wardana, Zuli Agustina Gultom — Mon, 31 Aug 2026 00:00:00 +0700

This study discusses the development of a web-based inventory prediction system for Toko Murni aimed at enhancing efficiency and accuracy in stock management. The main issue faced by Toko Murni was the inaccuracy of manual inventory management, which often resulted in discrepancies between records and actual stock levels, and failed to consider historical data patterns. The system was developed using PHP and MySQL, employing the Waterfall development model from requirement analysis to maintenance. Black Box testing was conducted to ensure all functions operate as expected. The results indicate that all features, including login, management of item types, period data, stock prediction, prediction results, password changes, and logout, function properly. The system accurately reflects inventory fluctuations and predicts stock needs quickly, supporting data-driven decision-making in a structured and effective manner. The simple yet functional interface facilitates administrators in managing inventory and recording data systematically.

SYSTEMATIC LITERATURE REVIEW PERKEMBANGAN METODE COMPUTER VISION PADA PENGOLAHAN CITRA TAHUN 2017–2025

Muhammad Arief Akbar, Fauzan Azim — Mon, 31 Aug 2026 00:00:00 +0700

The development of Computer Vision technology in recent years has shown significant growth along with advances in Deep Learning methods and the availability of large-scale datasets. Numerous studies have produced various approaches, architectures, and evaluation metrics, creating the need for a structured mapping to comprehensively understand the direction of this field. This study aims to analyze methodological trends, research task focuses, and dominant evaluation metrics in Computer Vision research. The method employed is a Systematic Literature Review (SLR) of 20 scientific article published between 2017 and 2025. The analysis process was conducted through data extraction covering method types, task focuses, datasets, and evaluation metrics used in each study. The results indicate that Convolutional Neural Networks and Vision Transformers are the most dominant architectures, with the primary research focuses on object detection, image classification, and video understanding. The most frequently used evaluation metrics are accuracy, mean Average Precision (mAP), and Intersection over Union (IoU). These findings reveal a gradual shift from convolution-based approaches toward transformer-based architectures that are more adaptive to large-scale visual data. This study provides a comprehensive overview of the development direction of Computer Vision and can serve as a reference for future research in selecting relevant methods and research focuses, both in terms of accuracy-oriented performance and computational efficiency.

PENERAPAN TEKNOLOGI REVERSE PROXY MENGGUNAKAN NGINX PADA INFRASTRUKTUR DIGITAL PEMERINTAH KOTA PEKANBARU

Dedy Yasriady; Asep Marzuki, Afifah Cahayani Adha — Mon, 31 Aug 2026 00:00:00 +0700

The rapid growth of web-based public services in local government has increased the complexity of managing web traffic, security, SSL certificates, and server administration across numerous subdomains. Without centralized management, these conditions can reduce operational efficiency and increase security risks. This study aims to evaluate the implementation of NGINX as a web gateway and reverse proxy within the digital infrastructure of the Pekanbaru City Government. A case study approach was employed through system design, implementation, and performance evaluation in a production environment managing the primary domain pekanbaru.go.id and approximately 170 subdomains. Data were collected from NGINX access logs, system monitoring using Grafana and Prometheus, and performance benchmarking using ApacheBench. The results show that the implementation of NGINX successfully centralized SSL certificate management, logging, and traffic routing through a single entry point, simplifying the administration of more than 170 subdomains while requiring only one public IP address. Benchmark testing indicates that although the average throughput of NGINX (163.07 requests/s) was slightly lower than the Apache backend (171.62 requests/s), NGINX demonstrated more stable performance with a significantly lower longest request time (2087 ms compared to 4039 ms) and no failed requests. These findings demonstrate that NGINX is an effective solution for improving operational efficiency, centralized security management, and service stability in government web infrastructures.

MITIGASI BIAS LARGE LANGUAGE MODEL MELALUI HUMAN-IN-THE-LOOP PADA AUTOMATED ESSAY SCORING BERBASIS ESAI IELTS-LIKE

Muhammad Abdiel Al Hafiz, M. Syaiful Amin — Mon, 31 Aug 2026 00:00:00 +0700

This study evaluates the reliability, algorithmic bias, and feasibility of implementing Human-in-the-Loop (HITL) in Large Language Model (LLM)-based Automated Essay Scoring (AES). Two architectures were compared: the Mixture-of-Experts architecture represented by GPT OSS 120B and the Dense architecture represented by Qwen3-32B, using a multiple-run scoring approach on 150 IELTS-like essays written by 50 respondents. Each essay was evaluated five times across four criteria: Grammar, Lexical Resource, Coherence, and Task Achievement. The evaluation employed the Intraclass Correlation Coefficient (ICC), Coefficient of Variation (CV), score range, paired t-test, and confidence-based routing simulation.The results show that GPT OSS 120B achieved higher reliability, with an ICC of 0.94, an average range of 2.40 points, and a CV of 4.69%, while Qwen3-32B obtained an ICC of 0.84, an average range of 5.56 points, and a CV of 9.17%. However, GPT OSS 120B experienced a parsing failure rate of 25.4%, whereas Qwen3-32B demonstrated full format compliance. Comparative analysis also showed that GPT OSS 120B tended to score more strictly, while Qwen3-32B was more lenient. In the Data Report task without visual input, both models exhibited conservatism bias due to contextual limitations. The HITL simulation showed that GPT OSS 120B could automatically approve 55.4% of essays, compared with only 7.8% for Qwen3-32B. These findings highlight the importance of HITL in maintaining the reliability, fairness, and integrity of academic assessment.

PENERAPAN FRAMEWORK ITIL DALAM MANAJEMEN LAYANAN TI PADA PLATFORM E-COMMERCE: LITERATURE REVIEW

Anisya RahmaNia, Aissah Meilia, Dwi Aprilia, Eriene Dheanda Absharina — Mon, 31 Aug 2026 00:00:00 +0700

The rapid growth of e-commerce has increased the need for effective information technology (IT) service management to ensure service availability, responsiveness, and user satisfaction. One of the widely adopted frameworks for managing IT services is the Information Technology Infrastructure Library (ITIL). This study aims to analyze the implementation of the ITIL framework in IT service management within e-commerce platforms through a literature review approach. The literature was collected from Google Scholar using keywords related to ITIL, IT Service Management (ITSM), and e-commerce. The selected studies were published between 2020 and 2025, available in full-text format, and sourced from scientific journals and conference proceedings. An initial search identified 30 articles, of which 7 met the inclusion criteria and were selected for further analysis. The findings indicate that ITIL V3 remains the most frequently adopted framework, while the use of ITIL V4 is still limited. Service Operation was identified as the most commonly applied domain, particularly in Incident Management, Problem Management, Request Fulfillment, and Access Management. Furthermore, most e-commerce platforms were found to operate at the Defined and Managed maturity levels. Common challenges reported across the studies include inadequate service monitoring, delayed incident resolution, system disruptions, limited service integration, and gaps in service availability and information security. Overall, the findings suggest that ITIL contributes to a more structured, measurable, and service-oriented approach to IT service management in e-commerce environments.

PERANCANGAN SISTEM INFORMASI POSYANDU BERBASIS WEBSITE SEBAGAI IMPLEMENTASI E-GOVERNMENT DESA MENGGUNAKAN DESIGN THINKING

Eka Firmani, Dwi Krisbiantoro, Oca Meilika Wulandari — Mon, 31 Aug 2026 00:00:00 +0700

Posyandu as a community-based health service still faces problems in data management carried out manually using conventional books, causing data to be prone to loss, unintegrated, and limited access to health information for the community. This condition hinders the effectiveness of health services for toddlers and the elderly at the village level. This study aims to design and develop a website-based Smart Health Village information system as an implementation of e-government at the village level using the Design Thinking approach. The Design Thinking method is applied through five stages, namely Empathize, Define, Ideate, Prototype, and Test, which enables system development based on the actual needs of users. The system development involved three posyandu cadres and six residents as informants. The resulting system has three user roles, namely Admin, Village Apparatus, and Residents, with main features including toddler and elderly data management, health examination recording, posyandu activity scheduling, and reports that can be exported to PDF format. System testing was carried out using the Black Box Testing method with twelve testing scenarios which showed that all functional features of the system ran as expected. With this system, it is hoped that posyandu health services can potentially improve service efficiency digitally, accurately, and sustainably as a concrete manifestation of e-government implementation at the village level.

PERANCANGAN WEBSITE MONITORING POSYANDU UNTUK DETEKSI DINI STUNTING DAN RESIKO PENYAKIT LANSIA

Oca Meilika Wulandari; Eka Firmani, Dwi Krisbiantoro — Mon, 31 Aug 2026 00:00:00 +0700

Healthcare service management at the village level is often hindered by manual and non-integrated recording systems, which limit early detection of health risks. This study aims to design a SmartHealth Village website using the Next.js framework to digitally monitor the health status of toddlers and the elderly, employing a Research and Development (R&D) method with a Waterfall model, conducted at Posyandu Desa Teluk, Purwokerto Selatan, Banyumas. Data were collected through observation and interviews with 9 informants consisting of 3 Posyandu cadres, 3 residents, and 3 village officials. The system automates the calculation of toddlers' nutritional status (stunting) based on WHO Z-Score standards and classifies hypertension and diabetes risk in the elderly based on JNC VII and Kemenkes RI (2020) guidelines. The website provides multi-level access for admin cadres, residents, and village officials, supported by a chatbot feature as a health education assistant. Functional testing using black-box testing showed that all 11 main features ran successfully. Performance evaluation indicated that data search time decreased from ±5–10 minutes to ±2–5 seconds, report generation time decreased from ±2–3 days to ±10–15 seconds, and calculation error rate decreased from ±10% to less than 1%, demonstrating that this platform is an efficient solution for transforming village-level healthcare management into a more transparent and responsive system.

PERANCANGAN SISTEM PENGARSIPAN ADMINISTRASI BERBASIS WEBSITE DI INSTITUT TEKNOLOGI DAN BISNIS MUHAMMADIYAH WAKATOBI

Putri Yapono, Asfin Herianto, Samsul Arif — Mon, 31 Aug 2026 00:00:00 +0700

At the Muhammadiyah Wakatobi Institute of Technology and Business, the administrative archive management process is still carried out manually, causing difficulties in document retrieval, the risk of archive loss, data duplication, and limited supervision of archive management activities. This study aims to design a Website-Based Administrative Archiving System that is able to manage administrative documents centrally and efficiently. The system development method used is the Software Development Life Cycle (SDLC) Waterfall model, with data collection techniques through observation, interviews, and literature studies. The system was built using HTML, Tailwind CSS, JavaScript, PHP, and MySQL, and modeled using the Unified Modeling language (UML). The resulting system includes archive upload features, document validation, search, statistical reports, activity logs, and role-based user management (role-based access control) for admins and users. Testing results using the Black Box Testing method show that all main functions of the system run according to functional requirements. Thus, the developed system is able to support centralized administrative archive management and is a solution to manual archiving problems at the Muhammadiyah Wakatobi Institute of Technology and Business

IMPLEMENTASI ALGORITMA GREEDY MATCHING PADA SISTEM REKOMENDASI JASA TUKANG BANGUNAN

Rosidin; Misbahus Sudur, Raditya Danar Dana, Ratna Dewi Lestyorini, Sukarsa — Mon, 31 Aug 2026 00:00:00 +0700

The small-scale construction industry in Warnasari District, Brebes Regency, still faces inefficiencies in finding daily construction workers because the recruitment process is commonly carried out through gathering points and word-of-mouth information. This condition creates income uncertainty for workers and makes it difficult for employers to assess workers’ skills, location, and reputation. This study aims to develop a web-based recommendation system for construction worker services by implementing the Greedy Matching algorithm. The main contribution of this research is the application of weighted matching based on skill, location distance, and performance rating criteria to generate fast and measurable worker recommendations. The system was developed using the Rapid Application Development (RAD) method, while the recommendation process was carried out by calculating the heuristic value of each candidate and selecting the candidate with the highest score as a local optimal solution. The evaluation was conducted through Black Box testing to assess the validity of system functions and a Greedy Matching calculation simulation to evaluate the candidate ranking output. The test results showed that all system features functioned validly with a success rate of 100%. The calculation simulation results showed that the candidate with a score of 97.0 was placed in the first position compared to another candidate with a score of 74.0. These results indicate that the system is able to generate measurable construction worker service recommendations based on the applied weighted criteria.

MACHINE LEARNING UNTUK PREDIKSI SUHU: SEBUAH TINJAUAN

Mon, 31 Aug 2026 00:00:00 +0700

Global climate change has increased the demand for accurate temperature prediction to support decision-making in sectors such as agriculture, disaster mitigation, and energy management. Machine Learning (ML) and Deep Learning (DL) approaches have been widely applied to model the non-linear and dynamic characteristics of temperature data. This study presents a Systematic Literature Review (SLR) following the PRISMA protocol. From 125 identified articles, 45 studies published between 2021 and 2025 were selected for detailed analysis. The results indicate that Long Short-Term Memory (LSTM) is the most frequently used algorithm, both as a standalone model and within hybrid architectures. Most studies employ multivariate datasets sourced from BMKG, ERA5 Reanalysis, satellite imagery, and the Internet of Things (IoT). Data preprocessing techniques, particularly norssmalization and time-series construction, play a crucial role in improving model stability. However, challenges remain, including hyperparameter sensitivity, complex weather data characteristics, and geographical variability. Future research opportunities include adaptive model development, multi-source data integration, and comprehensive comparative studies among algorithms.

ANALISIS PERFORMA ALGORITMA LIGHTGBM DENGAN TEKNIK SMOTE UNTUK KLASIFIKASI WEBSITE PHISHING

Angelyn Indriaty, Rizka Hafsari — Wed, 16 Sep 2026 00:00:00 +0700

Phishing website attacks are a severe cyber security threat that manipulates users into surrendering sensitive credentials. Machine learning detection based on URL lexical features offers a proactive, real-time alternative to reactive blacklists, but large-scale datasets such as URL-Phish 2025 (89.56% benign vs. 10.44% phishing across 111,660 records) exhibit extreme class imbalance that biases classifiers toward the majority class. This study analyzes the performance of combining the Synthetic Minority Over-sampling Technique (SMOTE) with the Light Gradient Boosting Machine (LightGBM) classifier, following the experimental baseline of Dam and Tran (2025). The dataset was divided using a stratified 75:10:15 split into training (83,745), validation (11,166), and testing (16,749) sets. SMOTE was applied only to the training set, balancing it to 75,000:75,000 samples in 0.5339 seconds. On 16,749 test samples, the optimal LightGBM + SMOTE model (Scenario II) achieved Accuracy of 97.94% (+4.29% over baseline), Recall of 98.23% (+56.83%), F1-Score of 90.88% (+33.25%), and ROC-AUC of 99.32%, while reducing False Negative threat leakage by 96.98% (from 1,025 to 31 cases), at an inference latency of only 0.0039 ms per URL. These results confirm that SMOTE and LightGBM together form an effective, real-time-ready cybersecurity detection solution.