International Journal on Human-Computing Studies

Optimization of Intelligent Systems for The Development of Algorithms Based on Graph Theory

2026-07-09T03:01:20+00:00

This research involves the study and analysis of information and data relating to enrolled and transient students on MOOC platforms, by applying graph theory to track students’ learning pathways and educational potential in order to identify their levels and difficulties, The study focuses on developing an applied scientific framework for use in unsupervised learning, utilising the Hidden Markov Model (HMM). This process benefits lecturers in scientific disciplines and supports intelligent systems that help students improve their practical and academic performance through the use of computers. In this case, we treat learning instances as observable events, and we can model the learning process using the HMM model.

Predicting Snack Food Production (Pentol) Using Fuzzy Logic

2026-07-18T10:47:28+00:00

Fluctuating market demand and limited inventory are challenges in determining the optimal amount of production, especially for small businesses such as meatball snack production. This study aims to determine the addition of meatball production using the Mamdani fuzzy logic method. This system uses three input variables: income, inventory, and sales demand, and one output in the form of the recommended amount of production. Each variable is modeled with a triangular membership function. The case study shows that with an income of Rp650,000, an inventory of 600 pieces, and a demand of 1,500 pieces, the fuzzy system recommends an additional production of 500 pieces. These results prove that the Mamdani fuzzy method is effective in helping production decision making amidst subjective or vague data uncertainty.

Development of the MY PERPUS Digital Library Website at SMA Muhammadiyah 4 Porong

2026-07-18T10:20:35+00:00

The rapid development of information technology has encouraged educational institutions to adopt digital-based systems to improve the effectiveness of administrative management, including school library services. The library of SMA Muhammadiyah 4 Porong still applies a manual system, which causes several problems such as recording errors, difficulties in searching for books, and inefficiency in the borrowing and returning process. This study aims to develop a website-based library information system called MY PERPUS to support library administration and improve service quality. The development method used in this research is the Waterfall model, which consists of requirement analysis, system design, and website development stages. Data were collected through observation and interviews with library staff and visitors. The developed system is equipped with login and registration features, book master data management, borrowing and returning transactions, and book return reminder notifications via Gmail. System testing using the black-box testing method indicates that all features function properly and produce outputs that meet the expected results. Therefore, the MY PERPUS website is able to improve the efficiency of library management, facilitate book searching, and help minimize delays in book returns..

Analysis of the Role of Artificial Intelligence (AI) and Blockchain Technology in Improving the Effectiveness of Fraud Detection Audits: A Literature Review

2026-07-18T11:07:17+00:00

The development of digital technology has transformed auditing practices, particularly in improving fraud detection effectiveness. The complexity of financial transactions, large data volumes, and limitations of conventional audit methods have encouraged the adoption of technologies such as Artificial Intelligence (AI) and Blockchain. This study analyzes the role of AI and Blockchain in enhancing audit effectiveness in fraud detection using a systematic literature review (SLR) approach. The method follows PRISMA guidelines by reviewing nationally and internationally indexed articles published between 2022 and 2024. The findings indicate that AI supports large-scale data analysis, anomaly detection, and predictive as well as real-time fraud identification. Meanwhile, Blockchain enhances audit data integrity, transparency, and reliability through an immutable and decentralized recording system. The integration of these technologies promotes a more effective and sustainable auditing system, although challenges remain regarding human resource readiness, technological infrastructure, and ethical and legal considerations.

Classification of Sentiment in Reddit Forum Comments using Fine-Tuned IndoBERT (Case Study: Free Nutritious Meal Program)

2026-07-18T10:35:14+00:00

This study analyzes public opinion on the Free Nutritious Meals Program (MBG) policy on the Reddit platform using the IndoBERT model. The study uses a dataset of 6,295 Reddit comments collected from 2024 to 2025. The preprocessing stage implemented a comprehensive suite of textual refinement procedures, encompassing normalization of colloquial language, emoji conversion, and translation into Indonesian to establish linguistic consistency across the corpus. Manual annotation was performed with meticulous care on a balanced dataset of 3,000 comments, evenly allocated among positive, negative, and neutral classes. An 80:20 train–test split was applied to enhance the reliability of model training and evaluation. A fine‑tuned IndoBERT model exhibited outstanding performance, attaining 99.00% for accuracy, F1‑score, and precision. Applied to the full dataset, the model predicted neutral sentiment as predominant (59.44%), followed by positive (34.11%) and negative (6.45%) sentiments, a distribution that suggests a measured and reflective public discourse on the topic. Model reliability was further supported by a mean confidence score of 0.8502 and a processing throughput of 137.93 samples per second, indicating strong potential for deployment as an effective tool for near real‑time sentiment monitoring in public policy contexts.

Interpretability-Driven Defense Against Physical Adversarial Attacks in Facial Recognition Systems

2026-09-19T11:04:11+00:00

This study aims to develop and evaluate a defense mechanism, the Self-Explaining Gate (SEG), to protect facial recognition (FR) systems against physical adversarial attacks (PAAs) by combining identity recognition, feature attribution, and anatomical landmark detection. SEG uses a three-stream architecture comprising identity feature extraction, real-time Grad-CAM feature attribution, and Haar-cascade-based anatomical landmark detection, integrated through a novel Anatomical Consistency Score (ACS) that measures the overlap between model attention and biological facial landmarks. The approach was evaluated on the CASIA-FASD and SiW benchmark datasets and a custom adversarial dataset, using detection accuracy, false acceptance/rejection rates, and inference latency as evaluation metrics. Our SEG achieved 94% attack detection accuracy, rejecting 96.1% of adversarial images while accepting 90% of genuine images, with inference latency under 160 milliseconds, outperforming adversarial training, feature squeezing, and SSR-FCN baseline defenses. Performance degrades on low-resolution images and under extreme occlusion, and further validation on video-based and larger real-world datasets is required. SEG can be integrated into real-time facial recognition deployments, such as border control, banking, and security systems, to provide interpretable and anatomically grounded rejection of adversarial inputs. By making automated identity verification more transparent and resistant to spoofing, SEG has the potential to increase public trust in facial recognition systems used in security-sensitive contexts. This study is the first to combine intrinsic, feedforward explainability with anatomical landmark validation as a real-time defense mechanism, introducing the Anatomical Consistency Score and a zero-background weighting scheme.

A Mutual Information-Driven Hybrid 1D-CNN and XGBoost Framework for Network Intrusion Detection in IoT

2026-10-03T01:07:41+00:00

There are a growing number of Internet of Things (IoT) interconnected systems in critical infrastructures that have made these systems vulnerable to increasingly advanced and diverse cyberattacks. However, typical intrusions systems face serious obstacles in operation due to the high dimensional telemetry data, inadequately extracting those non-linear structural patterns, and the inherent class imbalance in network traffic. To overcome these drawbacks, this paper presents an intelligent hybrid model based on Mutual Information (MI), One-Dimensional Convolutional Neural Network (1D-CNN) and eXtreme Gradient Boosting (XGBoost). First, redundant features are removed from the proposed pipeline by using an MI metric and the most discriminative flow attributes are selected by using an elbow heuristic. Later, an optimized 1D-CNN feature extractor is used to obtain non-linear representations, which are mapped to a compact deep feature manifold. Finally, a gradient boosted decision tree (GBDT) classifier is adapted to minimize the class imbalance and achieved a good separation between the benign and malicious flows. The proposed framework achieved an overall accuracy of 98.91%, an Area under the ROC Curve (AUC) of 0.9986, an intrusion class precision of 99.35%, an intrusion class recall of 99.27%, and an F1F_1-score of 99.31% by doing extensive empirical evaluations on the latest ToN-IoT benchmark dataset containing 37,979 independent test flow instances. Comparative analysis validates the effectiveness and reliability of this framework with the basic machine learning and deep learning approaches, giving a proper balance between discrimination and avoiding the majority class bias.