cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Much Aziz Muslim
Contact Email
a212muslim@mail.unnes.ac.id
Phone
+628164243462
Journal Mail Official
adminjoiser@shmpublisher.com
Editorial Address
SHM Publisher No. 64, Karanglo st, Pedurungan Distr, Semarang, Central Java, Indonesia 50191.
Location
Kota semarang,
Jawa tengah
INDONESIA
Journal of Information System Exploration and Research
Published by shm publisher
ISSN : 29641160     EISSN : 29636361     DOI : https://doi.org/10.52465/joiser.xxxx
Core Subject :
Journal of Information System Exploration and Research is a journal that publishes and disseminates scientific research papers on information systems to a wide audience particularly within the information system society
Arjuna Subject : -
Articles 13 Documents
 1   2 
Search results for , issue "vol. 4 no. 3: july 2026" : 13 Documents clear
Temperature, Humidity, and Weather Prediction Using Random Forest and LSTM for Food Crop Cultivation Optimization in Tegal Sarwo Edi; Aji Supriyanto
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.10

Abstract

The agricultural sector in the Tegal region faces uncertain climate fluctuations that directly impact food crop productivity. A crucial indicator for determining plant environmental comfort is the Temperature Humidity Index (THI). This research aims to develop a hybrid model capable of predicting and classifying future THI values to support precision decision-making for farmers. The methodology utilized historical climate data including Temperature humidity, rainfall, sunshine, and wind speed from the Tegal Maritime Meteorology Station spanning a 10-year period (2016-2025). A Long Short-Term Memory (LSTM) model was applied to forecast future THI values, while a Random Forest (RF) model was utilized to classify plant stress categories. Model performance was evaluated using Root Mean Square Error (RMSE), Accuracy, and F1-score. The results indicate that Tegal experiences comfortable (24 ≤ THI < 27), moderately comfortable (27 ≤ THI < 30), and uncomfortable (THI ≥ 30) conditions, particularly during dry and transitional seasons. The LSTM model achieved a 98.42% prediction accuracy, and the RF model reached a 99.59% classification accuracy. In conclusion, this highly accurate model can serve as an early warning system, providing farmers with actionable recommendations for optimal planting schedules and stress mitigation.
Approaching the Human Ceiling in Sleep Staging: A Controlled Comparison of Feature-Engineered and Deep Architectures Rand Nabeel Dawood
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.20

Abstract

Polysomnography (PSG) is the clinical gold standard for sleep assessment, yet manual epoch-by-epoch scoring is labor-intensive and subject to substantial inter-rater variability, which limits its use in large-scale, home-based, and routine clinical practice and makes reliable automated scoring an urgent need. This study suggests and compares an automated sleep-staging model based on two different supervised pipelines: a conventional feature-engineered Random Forest and a state-of-the-art end-to-end Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) network. Based on the recordings of the Sleep-EDF Expanded database, we preprocessed the multichannel signals (EEG and EOG) into 30-second epochs, which were classified into five stages (Wake, N1, N2, N3 and REM). Our findings show that the feature-engineered baseline is very competitive with 95.49% accuracy, but the CNN-LSTM model performs better (96.44% accuracy) and has a much higher capability of classifying the transitional N1 stage which is harder. Because these results are obtained from a single overnight recording with a within-subject train/test split, they are best interpreted as a controlled proof-of-concept upper bound rather than as evidence of clinical-grade or human-level performance; cross-subject (leave-one-subject-out) validation on multiple recordings is required before any such claim can be made.
Development of a Data Analytics-Based Decision Support System for Optimizing Public Transportation Services Maria Atik Sunarti Ekowati Maria; Nurul Hidayat
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.73

Abstract

Public transport is essential for sustainable urban mobility, but challenges such as limited capacity, operational inefficiency, and service performance remain. This study develops a Decision Support System (DSS) that integrates operational data, passenger preferences, and performance indicators to provide recommendations for decision-making. The study uses passenger surveys, system performance analysis, and data analysis and prediction techniques. The DSS incorporates information visualization, performance monitoring, and scenario analysis to help decision-makers anticipate potential outcomes and make informed choices. The results demonstrate improvements in service performance, with passenger satisfaction increasing from 76% to 95%, while prediction accuracy exceeded 90%. These findings indicate that the DSS can support more effective decision-making and resource allocation while improving transparency and accountability in public transport services. The system also aligns with efforts to promote technology-driven urban development. Overall, integrating data and decision-support technologies can improve public transport operations, support better resource utilization, and contribute to more sustainable and inclusive urban mobility.
SentinelKEM: Securing AI Model Distribution Using Hybrid Cryptography Singgih Naufal; Eko Hari Rachmawanto; Mohamed Doheir
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.76

Abstract

Deep neural network (DNN) models are valuable intellectual property that can be copied or redistributed without authorization during distribution. Existing protection methods mainly rely on watermarking, which verifies ownership but does not secure model files during transfer. This study proposes SentinelKEM, a hybrid cryptographic framework that combines AES-256-GCM for authenticated encryption, Kyber/ML-KEM-768 with HKDF-derived key encryption keys for post-quantum key protection, Ed25519 digital signatures for ownership verification, and scrypt for passphrase-based private-key security. Encrypted models and cryptographic metadata are packaged into a secure ZIP archive, while private keys are distributed separately. During decryption, SHA-256 integrity verification and Ed25519 authentication are performed before model restoration. SentinelKEM was implemented as a Python Streamlit application and evaluated on various AI model formats, including .h5, .pt, .pth, .pkl, and .joblib. Experimental results showed successful encryption and decryption, reliable ownership authentication, effective tamper detection, and a constant cryptographic overhead of approximately 4.5 KB regardless of model size. Unlike watermarking, SentinelKEM protects AI models before recipient access through post-quantum key encapsulation and authenticated encryption, providing practical and robust security for AI model distribution.
Explainable Ensemble Learning for Identifying Digital Citizenship Drivers Among Generative AI Users: A SHAP-Based Behavioral Analysis Muchammad Thoha; Andy Prasetyo Utomo; Zainur Romadhon
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.79

Abstract

The increasing integration of Generative Artificial Intelligence (Generative AI) into higher education has transformed how university students learn and interact within digital environments. However, the psychological and behavioral factors influencing responsible Digital Citizenship behavior among Generative AI users remain insufficiently understood. Existing studies have largely relied on descriptive, regression-based, or structural modeling approaches that may overlook complex nonlinear behavioral relationships. This study proposes an explainable ensemble learning framework to identify the dominant drivers of Digital Citizenship among university students using Generative AI technologies. A national-scale dataset consisting of 2,008 students from 41 universities across 22 provinces in Indonesia was analyzed using Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Support Vector Regression (SVR). Model performance was evaluated using R², MAE, RMSE, and cross-validation procedures. To improve transparency and interpretability, SHAP (Shapley Additive Explanations) was employed to examine feature importance and nonlinear behavioral effects. The results showed that XGBoost achieved the highest predictive performance. SHAP analysis consistently identified technology anxiety, self-efficacy, and attitude toward behavior as the most influential predictors of Digital Citizenship. The findings further revealed nonlinear effects of technology anxiety on responsible AI participation. This study contributes to explainable educational analytics by providing interpretable insights into responsible AI behavior and supporting evidence-based AI governance and digital education policy in higher education.
Machine Learning-Based Obesity Prediction and Feature Importance Analysis Using Random Forest Anjelina Dwi Puspita; Nelly Astuti Hasibuan; Kevin Boy Sinaga; Krisna Apta Jaya Zendrato; Muhammad Alief Rahman Susanto
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.82

Abstract

Obesity prevalence continues to rise both globally and in Indonesia, yet most predictive studies rely on datasets from Western populations or clinical settings, leaving a gap in machine-learning-based evidence from Southeast Asian university communities. This study applies Random Forest to survey data from 300 respondents in Medan, Indonesia, to identify dominant determinants of Body Mass Index (BMI) category and to evaluate predictive reliability under class imbalance and limited sample size. After cleaning, 297 valid records remained; twelve behavioral and demographic features were used as model inputs. Beyond a conventional 80:20 train-test split, this study applied SMOTE to address minority-class imbalance and 10-fold stratified cross-validation to assess stability. The 80:20 split yielded 83.33% accuracy (weighted F1 = 0.82); 10-fold cross-validation produced a more conservative mean accuracy of 77.05% (SD = 6.37%), confirming that single-split evaluation overstated performance. SMOTE improved minority-class (Obesity) recall from 0.67 to 0.78 without reducing overall accuracy. Age emerged as the dominant predictor (importance = 0.25), followed by meal frequency, physical activity, and vegetable consumption frequency. These findings support Random Forest as a viable obesity-risk screening tool in resource-limited, questionnaire-based settings, while highlighting the need for imbalance-aware evaluation on modest, real-world health-survey samples. 
Comparison of Shallow and Deep Learning for Indonesian Clickbait Headline Classification Muhammad Noer Attalah Dzahkwan; Majid Rahardi
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.85

Abstract

Clickbait is an increasingly prevalent phenomenon in Indonesian online news media, where headlines are crafted to attract clicks without accurately reflecting article content. This study proposes and compares eight classification models: three shallow learning algorithms — Naive Bayes, Support Vector Machine (SVM), and Logistic Regression — and five transformer-based deep learning models: IndoBERT-p1, IndoBERT-p2, XLM-RoBERTa, mBERT, and DistilBERT. The dataset used is CLICK-ID, consisting of 15,000 labeled headlines from 12 Indonesian news portals, expanded to 25,138 samples via semi-supervised pseudo labeling with a confidence threshold of 0.85. All deep learning models were trained with Focal Loss (α=0.25, γ=2.0) to address class imbalance and Automatic Mixed Precision (AMP) for GPU efficiency. Results show that IndoBERT-p1, IndoBERT-p2, XLM-RoBERTa, mBERT, and DistilBERT achieve comparable performance, with macro F1-scores ranging from 86.57% to 88.54%. Among shallow learning models, SVM performs best with 83.51% F1-score. An average ensemble of all five transformer models achieves the best overall performance at 90.14% accuracy and 89.00% F1-score, outperforming every individual model. This study contributes to Indonesian clickbait detection research by demonstrating that ensemble aggregation of diverse transformer architectures yields more reliable performance than reliance on any single model.
Investigating Mobile Legends Top Up Fraud on Instagram Using the National Institute of Justice Method Eko Aribowo; Auliana
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.88

Abstract

The rapid growth of digital technology has increased the occurrence of cybercrime, including fraud through social media platforms. This study investigates a Mobile Legends top-up fraud case carried out via Instagram using the National Institute of Justice (NIJ) digital forensic framework. The NIJ method consists of five stages: identification, collection, examination, analysis, and reporting. The primary data source was an Instagram archive in JSON format obtained through the Download Your Information feature and examined using Oxygen Forensic Detective. Evidence integrity was verified using FTK Imager through hash verification, while Metadata2Go was utilized for metadata analysis. The investigation successfully recovered 51 digital artifacts, comprising 1 Instagram account profile, 3 promotional posts, 2 testimonial highlights, 19 Direct Message (DM) communication threads, and 15 digital transfer receipts, identifying 15 distinct fraud victims. Hash verification yielded 100% match accuracy (MD5 and SHA-1), validating that the evidence remained unaltered. The findings revealed a fraud pattern characterized by payment acceptance without service fulfillment, followed by communication termination. Furthermore, metadata analysis and hash verification confirmed that the digital evidence was authentic, unaltered, and forensically reliable. These results demonstrate that Instagram archive data can serve as valid digital evidence in social media fraud investigations.
Smart Expert System for Tuberculosis Diagnosis Using the Naïve Bayes Method Ulumuddin Ulumuddin; Siti Harlina; Warjiyono; Amin Nur Rais; Arham Arifin; Abdul Ibrahim; James Adam Seo
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.93

Abstract

Tuberculosis (TB) remains one of the major infectious diseases worldwide, requiring accurate early diagnosis to reduce transmission and improve treatment outcomes. Although numerous machine learning-based diagnostic systems have been developed, most previous studies have primarily focused on improving classification accuracy without integrating an expert system to support clinical decision-making. This study aims to develop a Smart Expert System for the early diagnosis of tuberculosis using the Naïve Bayes algorithm. The novelty of this research lies in the integration of the Naïve Bayes algorithm with a web-based expert system that not only generates diagnostic predictions but also provides recommendation-based decision support. The proposed system was developed using 200 tuberculosis patient records consisting of demographic data and clinical symptoms. The research process included data preprocessing, model training, system implementation, and performance evaluation using Accuracy, Precision, Recall, and F1-score metrics. The experimental results achieved an Accuracy of 87.00%, Precision of 85.00%, Recall of 88.00%, and an F1-score of 86.00%, indicating reliable classification performance. The developed system assists healthcare professionals in conducting preliminary TB screening, accelerating clinical decision-making, and improving public access to early tuberculosis diagnostic information.
Predicting Student Academic Performance Using the Naïve Bayes Algorithm (Case Study: SMK Pelayaran “AKPELNI” Semarang) Aditya Desta Saputra; Herny Februariyanti
Journal of Information System Exploration and Research Vol. 4 No. 3: July 2026
Publisher : shmpublisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52465/joiser.v4i3.95

Abstract

Predicting student academic performance can help schools identify students who may require early academic support. This study evaluates a Naïve Bayes classification approach using attendance and parental socioeconomic factors to predict academic performance at SMK Pelayaran “AKPELNI” Semarang. The study used 210 student records from the 2024/2025 academic year. Academic performance was determined from the average scores in Indonesian Language, Mathematics, English, and vocational subjects and classified into Low, Medium, and High categories. These subject scores were excluded as predictors to maintain independence between the predictors and the target label. The predictors consisted of total attendance, father’s education and income, and mother’s education and income. Missing parental education values were handled using mode imputation. The data were divided using a stratified 80:20 train-test split. Categorical Naïve Bayes with Laplace smoothing (α=1) and uniform class priors was applied. On the 42-record test set, the model achieved 66.67% accuracy, 39.74% balanced accuracy, and 33.88% macro F1. The Medium class achieved an F1 score of 79.41%, the High class 22.22%, while the Low class was not detected. These findings indicate that attendance and parental socioeconomic factors alone are insufficient for reliable minority-class prediction and should not yet be used as a standalone early warning system.

Page 1 of 2 | Total Record : 13

 1   2 

Filter by Year

2026 2026


Reset
Filter By Issues
All Issue Vol. 4 No. 4: October 2026 Vol. 4 No. 3: July 2026 Vol. 4 No. 2: April 2026