Jurnal Teknik Informatika (JUTIF)
Jurnal Teknik Informatika (JUTIF) is an Indonesian national journal, publishes high-quality research papers in the broad field of Informatics, Information Systems and Computer Science, which encompasses software engineering, information system development, computer systems, computer network, algorithms and computation, and social impact of information and telecommunication technology. Jurnal Teknik Informatika (JUTIF) is published by Informatics Department, Universitas Jenderal Soedirman twice a year, in June and December. All submissions are double-blind reviewed by peer reviewers. All papers must be submitted in BAHASA INDONESIA. JUTIF has P-ISSN : 2723-3863 and E-ISSN : 2723-3871. The journal accepts scientific research articles, review articles, and final project reports from the following fields : Computer systems organization : Computer architecture, embedded system, real-time computing 1. Networks : Network architecture, network protocol, network components, network performance evaluation, network service 2. Security : Cryptography, security services, intrusion detection system, hardware security, network security, information security, application security 3. Software organization : Interpreter, Middleware, Virtual machine, Operating system, Software quality 4. Software notations and tools : Programming paradigm, Programming language, Domain-specific language, Modeling language, Software framework, Integrated development environment 5. Software development : Software development process, Requirements analysis, Software design, Software construction, Software deployment, Software maintenance, Programming team, Open-source model 6. Theory of computation : Model of computation, Computational complexity 7. Algorithms : Algorithm design, Analysis of algorithms 8. Mathematics of computing : Discrete mathematics, Mathematical software, Information theory 9. Information systems : Database management system, Information storage systems, Enterprise information system, Social information systems, Geographic information system, Decision support system, Process control system, Multimedia information system, Data mining, Digital library, Computing platform, Digital marketing, World Wide Web, Information retrieval Human-computer interaction, Interaction design, Social computing, Ubiquitous computing, Visualization, Accessibility 10. Concurrency : Concurrent computing, Parallel computing, Distributed computing 11. Artificial intelligence : Natural language processing, Knowledge representation and reasoning, Computer vision, Automated planning and scheduling, Search methodology, Control method, Philosophy of artificial intelligence, Distributed artificial intelligence 12. Machine learning : Supervised learning, Unsupervised learning, Reinforcement learning, Multi-task learning 13. Graphics : Animation, Rendering, Image manipulation, Graphics processing unit, Mixed reality, Virtual reality, Image compression, Solid modeling 14. Applied computing : E-commerce, Enterprise software, Electronic publishing, Cyberwarfare, Electronic voting, Video game, Word processing, Operations research, Educational technology, Document management.
Articles
1,242 Documents
<![CDATA[Indonesian Hate Speech Detection: A Cross-Validated Benchmark of Machine Learning and Pre-trained Transformer Models with Statistical Significance Analysis ]]>
<![CDATA[Dodo Zaenal Abidin]]>;
<![CDATA[Agus Siswanto]]>;
<![CDATA[Chindra Saputra]]>;
<![CDATA[Imelda Yose]]>;
<![CDATA[Kaslin Kaslin]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5176
<![CDATA[Automated hate speech detection in Indonesian social media remains a persistent challenge due to dataset fragmentation, heterogeneous annotation schemes, and the lack of reproducible cross-model benchmarks with formal statistical validation. This study presents a cross-validated benchmark that systematically evaluates seven models — one majority baseline, three classical machine learning models (Logistic Regression, SVM, Random Forest), and three transformer-based pre-trained language models (DistilBERT, IndoRoBERTa, IndoBERT) — on a standardized multi-source Indonesian hate speech corpus comprising 14,043 samples from Twitter and Instagram. All models were trained and evaluated under identical conditions using stratified three-way splits (70/15/15) replicated across three random seeds, reporting mean ± standard deviation for F1-macro, precision-macro, recall-macro, and ROC-AUC. Paired t-tests and Cohen's d effect size analysis were applied to formally assess statistical significance and practical magnitude of performance differences. Results show that all transformer-based models significantly outperformed all classical ML models, with IndoBERT achieving the highest mean F1-macro of 0.8834 (±0.0026) and the lowest cross-seed variance among all models. Notably, IndoBERT and IndoRoBERTa were found to be statistically equivalent (p=0.7273, d=0.283), indicating that neither model is definitively superior for this task. Among classical models, SVM attained the best F1-macro of 0.8509. These findings confirm that domain-specific pre-training on Indonesian corpora contributes to both higher performance and superior cross-seed stability. The proposed benchmarking framework, standardized corpus, and statistical evaluation protocol provide a reproducible reference for future Indonesian hate speech detection research, thereby advancing the methodological standards of automated text classification in computer science and informatics, particularly for under-resourced language NLP research.]]>
<![CDATA[Lightweight Early and Late Blight Detection on Potato Leaves via Knowledge Distillation for Precision Agriculture ]]>
<![CDATA[Ahmad Faisol]]>;
<![CDATA[Deddy Rudhistiar]]>;
<![CDATA[Betty Dewi Puspasari]]>;
<![CDATA[Thesa Adi Saputra Yusri]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5181
<![CDATA[Accurate and timely detection of plant leaf diseases plays a vital role in ensuring crop health and supporting sustainable agricultural practices, particularly in the context of food security. Potato (Solanum tuberosum) is a high-potential food crop and a strategic commodity in many countries, including Indonesia, due to its nutritional value and adaptability to various agro-climatic conditions. However, its productivity is highly vulnerable to diseases such as early blight and late blight. This study presents a knowledge distillation framework for developing an efficient deep learning model to classify potato leaf diseases. EfficientNet-B5 was employed as the teacher model, achieving 100% accuracy, F1-score, Matthews Correlation Coefficient (MCC), and Cohen’s Kappa on the validation set. The student model, based on MobileNetV3-Small, successfully retained high predictive performance, achieving 99.07% accuracy, a macro F1-score of 0.9828, and a Cohen’s Kappa of 0.9833. MobileNetV3-Small significantly improved efficiency by reducing inference time by 67.58% (from 33.44 ms to 10.84 ms) and model size by 96.51% (from 111.54 MB to 3.89 MB) compared to EfficientNet-B5, making it highly suitable for real-time and resource-constrained applications. These results confirm that knowledge distillation enables the construction of lightweight models without significant loss of accuracy, making them suitable for mobile and edge-based agricultural applications.]]>
<![CDATA[IndoBERT-based Named Entity Recognition Using Transformer Model for Indonesian Waste Bank Data Processing ]]>
<![CDATA[Mardiyyah Hasnawi]]>;
<![CDATA[Wistiani Astuti]]>;
<![CDATA[Andi Puspitasari]]>;
<![CDATA[Nia Kurniati]]>;
<![CDATA[Dolly Indra]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5215
<![CDATA[Data processing in waste bank management systems faces substantial challenges when extracting structured information from unstructured textual data containing transaction records, customer information, and waste categorization details. This study develops and validates an automated Named Entity Recognition (NER) system based on the IndoBERT transformer architecture to process textual records from Makassar waste bank operations. The approach fine-tunes the indolem/indobert-base-uncased model through domain-specific adaptation targeting waste management vocabulary and operational terminology. The dataset comprises 975 textual records collected from 8 Waste Bank Units in Makassar City. These records were systematically annotated using the BIO tagging scheme for eight entity types: B-LOCATION, B-PERSON, B-WASTE_CATEGORY, B-WASTE_TYPE, I-PERSON, I-WASTE_CATEGORY, I-WASTE_TYPE, and O. The dataset was partitioned into training (682 samples, 69.9%), validation (146 samples, 15.0%), and test sets (147 samples, 15.1%) using stratified sampling methodology with high inter-annotator agreement (κ=0.91). Experimental results show outstanding performance with 94.37% test accuracy, 90.65% precision, 94.37% recall, and 92.21% F1-score, outperforming general-purpose Indonesian NER approaches. Perfect performance was achieved for waste type recognition (B-WASTE_TYPE: 100% F1-score) and location identification (B-LOCATION: 100% F1-score), while waste category classification reached 96% F1-score. This implementation successfully automates entity extraction from Makassar waste bank textual data, reducing manual processing time by 95% while maintaining high accuracy levels. This research makes important contributions to Indonesian environmental natural language processing. These contributions include transformer adaptation methodologies for resource-constrained domains, validated IndoBERT performance on Indonesian waste bank data, the first Indonesian waste management NER dataset, and demonstrated feasibility of NLP-based environmental policy systems.]]>
<![CDATA[Detecting And Classifying Multi-Label Semantic Bias In 3,829 Indonesian Military Criminal Judgments Dataset Using Language Modeling And Ensemble Strategies ]]>
<![CDATA[Bayu Ardiyansyah]]>;
<![CDATA[Lutfi Indra Nur Praditya]]>;
<![CDATA[Galih Wasis Wicaksono]]>;
<![CDATA[Nur Putri Hidayah]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5372
<![CDATA[Objectivity in military criminal judgments is crucial for judicial legitimacy but is frequently compromised by semantic bias. To the best of our knowledge, this is the first study to specifically address automated bias detection within the Indonesian military legal domain, bridging a significant gap in the literature that has predominantly focused on general civil law. This study aims to develop a multi-label classification model to automatically detect and classify three specific types of bias (emotional, character, and ambiguity) in military legal texts. The methodology involved the acquisition and expert annotation of 3,829 judgment documents (2020–2025). Three feature extraction strategies (TF-IDF, IndoBERT, and Doc2Vec) were comparatively evaluated using KNN, MLP, Random Forest, and Custom Ensemble algorithms. Experimental results demonstrate that the lexical approach using Random Forest with TF-IDF achieved superior performance with a weighted F1-Score of 0.82, outperforming both complex embedding-based models and the ensemble approach (F1-Score 0.77). The findings further reveal that character bias is the most dominant form of distortion in the corpus. This research makes three novel contributions: (1) providing the first annotated legal dataset for the Indonesian military domain; (2) demonstrating the superior efficacy of lexical features (TF-IDF) over complex embeddings in this specific legal domain; and (3) establishing a technical foundation for a decision-support system to enhance judicial objectivity.]]>
<![CDATA[Comparative Analysis of Discord Artifact Recovery from Browser Cache Using Improved Generic Computer Forensic Investigation Model (GCFIM) ]]>
<![CDATA[Mikhail Ashshidiqie Rachman]]>;
<![CDATA[Yudi Prayudi]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5377
<![CDATA[The rapid growth of Discord as a web-based communication platform has increased its relevance in cybercrime investigations, particularly in cases involving deleted digital evidence. Recovering residual artifacts from browser cache is therefore essential to support digital forensic analysis. This study aims to compare the effectiveness of forensic tools in recovering Discord artifacts from browser cache across Google Chrome, Mozilla Firefox, and Microsoft Edge. A live forensic approach was applied using the Improved Generic Computer Forensic Investigation Model (Improved GCFIM) to ensure a systematic and reproducible investigation process. Experiments were conducted by simulating identical Discord user interactions on each browser, including text messaging, emoji and sticker usage, and media file transfers, followed by artifact extraction using Autopsy, ChromeCacheView, and MZCacheView.The experimental results show that Mozilla Firefox produced the highest volume of recoverable data, with 484 cache files identified, allowing successful artifact extraction using both Autopsy and MZCacheView. Google Chrome and Microsoft Edge yielded 146 and 275 cache files respectively, where artifact recovery was only effective using ChromeCacheView due to their Chromium-based cache structure. Across all browsers, recovered artifacts included user identities, channel identifiers, timestamps, text messages (plain, edited, pinned, spoiler, and reply), emojis, stickers, images, videos, and documents, enabling accurate reconstruction of communication timelines. These findings demonstrate that browser architecture and tool compatibility significantly influence cache-based evidence recovery and highlight the necessity of a multi-tool, browser-aware forensic approach. This research contributes to digital forensics and computer science by strengthening cache-based evidence recovery and providing practical insights into browser-specific forensic analysis of web-based communication platforms.]]>
<![CDATA[Biometric Authentication Improving Robustness Attendance System with MobileNet ]]>
<![CDATA[Muhammad Imanullah]]>;
<![CDATA[Sandhy Fernandez]]>;
<![CDATA[Ardi Wijaya]]>;
<![CDATA[M. Yoka Fathoni]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5381
<![CDATA[Our research is focused on developing an effective attendance system that matches the conditions of pandemic or other specific pandemic situations. We have built a legacy system using Randomized QR-Code and MAC-Address as the authentication method. Based on the analysis, it is known that users need around 25.8877 seconds to authenticate their presence in the system. Therefore, we are trying to improve the authentication time and robustness of the system by using user facial biometrics. Facial biometrics was chosen because it is the most appropriate way to authenticate one's identity during pandemic, where physical contact is unnecessary or even expected to be avoided. According to several test scenarios, it is known that our new system takes time simultaneously to recognize the user and record their presence data once they show their face to the recognition camera. This new system can also run more powerfully with a percentage of 1.26% processor time, 16.9059% faster than the previous system, because it is deployed in a web browser using the latest web app framework technology, which can run on various devices. Based on these findings, our new approach successfully improves its predecessor systems. To produce a robust attendance system, we must implement a Biometric Authentication method, which is enhanced with other secure authentication methods such as Random QR-Code scanning. With a combination of these methods, an attendance system will run safely and efficiently.]]>
<![CDATA[Implementation of Position Weighted Retrieval BM25 (PoWeR-BM25) for Thesis Recommendation Systems ]]>
<![CDATA[Putu Agung Ananta Wijaya]]>;
<![CDATA[I Made Gede Sunarya]]>;
<![CDATA[I Gede Aris Gunadi]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5414
<![CDATA[The increasing volume of academic documents in University of Udayana Digital Library (e-Perpus UNUD) has increased the difficulty for students to efficiently identify relevant prior research aligned with their research topics. Conventional keyword-based search mechanisms often fail to capture contextual relevance, resulting in information overload and suboptimal retrieval quality. To address this challenge, this study presents the design and implementation of a thesis recommendation system based on a modified BM25 algorithm called Position-Weighted Retrieval BM25 (PoWeR-BM25). The proposed model introduces an adaptive positional weighting scheme that emphasizes query term occurrences appearing at the beginning and end of a document, where essential contextual information is typically concentrated. The system was developed using an information retrieval framework and evaluated across multiple configurations combining different data representations (title, abstract, and title–abstract) and preprocessing techniques (stemming and non-stemming). Experimental results demonstrate that POWER-BM25 consistently outperforms traditional BM25 and TF-IDF in terms of Mean Average Precision (MAP) and F1-Score, particularly when applied to stemmed abstract data. The best performing configuration was subsequently deployed as recommendation feature integrated into University of Udayana Digital Library via a Web Service API, enabling users to obtain more accurate and contextually relevant thesis recommendations. These findings highlight the practical importance of incorporating positional term weighting into probabilistic retrieval models as a lightweight yet effective approach to improving academic recommendation systems without relying on computationally intensive machine learning models.]]>
<![CDATA[Feature Selection Using Genetic Algorithm Combined with Linear Discriminant Analysis and k-Nearest Neighbors for High-Dimensional Text Classification ]]>
<![CDATA[Eko Puji Laksono]]>;
<![CDATA[M Zainal Arifin]]>;
<![CDATA[Aji Prasetya Wibawa]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5423
<![CDATA[Text classification is an important task in natural language processing that faces the main challenge of high dimensionality in the representation of text features, such as Bag-of-Words and TF-IDF. High dimensionality leads to data sparsity, increased computational complexity, and decreased model performance due to irrelevant and redundant features. Therefore, an effective feature selection method is needed to improve the accuracy and efficiency of text classification. This study aims to develop an optimal feature selection method by integrating Genetic Algorithm (GA) as a wrapper method, Linear Discriminant Analysis (LDA) for dimension reduction, and k-Nearest Neighbors (k-NN) as a classifier in a high-dimensional text classification pipeline. GA is used to explore a subset of features in an adaptive and global manner, LDA transforms features into spaces with maximum class separability, and k-NN classifies based on proximity within optimized feature spaces. Experiments were conducted on three standard text datasets, namely 20 Newsgroups, Reuters-21578, and BBC News, with a comprehensive evaluation using a variety of preprocessing configurations and feature representations. The results showed that the GA+LDA+k-NN approach significantly improved the classification accuracy compared to traditional feature selection methods such as Chi-Square, Information Gain, and Mutual Information. This approach is also more robust against noise and is able to handle minority classes better. These findings confirm the urgency of using evolutionary optimization-based feature selection methods in high-dimensional text classification, which not only improves model performance but also provides a practical solution for large-scale and complex text data processing. Thus, this research makes an important contribution both theoretically and applicatively in the development of modern text classification systems. ]]>
<![CDATA[Content-Based Recommendation System for Non-Textbook using TF-IDF, Cosine Similarity, and Educational Level Filtering ]]>
<![CDATA[Arif Rohmadi]]>;
<![CDATA[Ristu Saptono]]>;
<![CDATA[Brilyan Hendrasuryawan]]>;
<![CDATA[Bambang Widoyono]]>;
<![CDATA[Akhmad Syaifuddin]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5425
<![CDATA[Non-textbook educational resources refer to books designed to enrich readers’ knowledge, insights, and skills, serving as complementary materials to formal textbooks. These may include fiction, non-fiction, biographies, self-help books, and other supplementary literature. However, the broad range of available non-textbooks targeting diverse educational levels often presents challenges for students in selecting materials that are appropriate to their academic stage. This study aims to develop a content-based recommendation system capable of recommending non-textbooks based on the reader's educational level. The recommendation process employs Term Frequency–Inverse Document Frequency (TF-IDF) for feature extraction and Cosine Similarity to calculate semantic similarity between user search queries and book content. To improve relevance, a filtering mechanism based on education level is introduced prior to feature extraction. Experimental results show that applying this filtering process significantly improves the recommendation's performance, yielding an average precision of 100%. In contrast, models without the filtering process achieve only 50% precision. These findings highlight the effectiveness of contextual filtering in improving the accuracy of non-textbook recommendation systems. ]]>
<![CDATA[Hybrid MQTT–AMQP–SSE Architecture for Efficient Data Delivery in Multi-Sensor IoT Environmental Monitoring: A Topology Comparison Study ]]>
<![CDATA[Iin Karmila Yusri]]>;
<![CDATA[Agus Purnawan]]>;
<![CDATA[Ismail Yusuf Panessai]]>
<![CDATA[ Jurnal Teknik Informatika (Jutif) Vol. 7 No. 4 (2026): JUTIF Volume 7, Number 4, August 2026 ]]>
Publisher : <![CDATA[Informatika, Universitas Jenderal Soedirman ]]>
Show Abstract
|
Download Original
|
Original Source
|
Check in Google Scholar
|
DOI: 10.52436/1.jutif.2026.7.4.5438
<![CDATA[The rapid growth of the Internet of Things (IoT) has created increasing demands for reliable and efficient data delivery mechanisms in wireless multi-sensor systems, particularly in terms of latency, scalability, and synchronization. This work introduces an efficient data transmission architecture for a wireless multi-sensor monitoring system designed for IoT-based environmental applications. The suggested design incorporates ESP8266-based sensor nodes, RabbitMQ as a message broker, InfluxDB for the storage of time-series data, and a Flask-based web interface for the viewing of real-time data. A hybrid communication protocol integrating MQTT, AMQP, and Server-Sent Events (SSE) ensures rapid data transmission, accurate message queuing, and continuous online streaming availability. We developed and evaluated three network topologies: Star, Tree, and mesh. We employed them to assess latency, throughput, packet loss, and synchronization precision. The Tree topology is optimal, exhibiting a latency of 2 to 4 seconds, ensuring 100% data consistency, and maintaining reliable communication within an 80-meter radius. The RabbitMQ broker exhibits minimal latency (30–272 ms) and has no packet loss, even under high message volume. Wi-Fi mode 802.11 N exhibits superior signal range and stability compared to mode B. The findings indicate that the integration of gateway-based aggregation with a hybrid multi-protocol communication model enhances data reliability and scalability. This provides useful insights for data transfer and distributed data management in IoT monitoring systems based on informatics.]]>
