Garuda - Garba Rujukan Digital

Analisis Sentimen untuk Ulasan Produk E-Commerce Shopee Menggunakan BERT Sikana, Nadya; Winardi, Sunaryo; -, Gunawan; Situmorang, Gilbert Fernando; Lubis, Rivaldi
Jurnal Sifo Mikroskil Vol. 26 No. 2 (2025): JSM VOLUME 26 NOMOR 2 TAHUN 2025
Publisher : Fakultas Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55601/jsm.v26i2.1796

Analisis sentimen sangat penting untuk memahami opini konsumen dan menyempurnakan strategi e-commerce. Analisis ini menghadapi tantangan seperti bahasa informal, ambiguitas semantik, dan inkonsistensi antara sentimen tekstual dan peringkat bintang, yang memengaruhi akurasi klasifikasi. Penelitian ini menerapkan model BERT (Bidirectional Encoder Representations from Transformersi) untuk mengklasifikasikan sentimen dalam ulasan pengguna Shopee. Data dikumpulkan dari penelitian sebelumnya dan menjalani praproses, termasuk tokenisasi, penghapusan stopword, dan normalisasi teks. Pendekatan analisis sentimen berbasis leksikon digunakan sebagai dasar perbandingan. Model BERT disempurnakan menggunakan optimasi hiperparameter, mencapai akurasi 83,08%, presisi 82,91%, recall 83,08%, dan F1-score 82,87%. Dibandingkan dengan studi sebelumnya yang menggunakan Naïve Bayes dengan N-Gram dan Information Gain, yang mencapai akurasi 92% tetapi presisi lebih rendah (56%), recall (65%), dan F1-score (60%), BERT mengungguli dengan metrik evaluasi yang lebih seimbang dan keandalan prediktif yang lebih besar. Hasil ini menunjukkan kemampuan BERT untuk menangkap konteks semantik dua arah, melampaui metode tradisional dalam menangani tugas analisis sentimen yang kompleks.

LITERATURE REVIEW OF THE APPLICATION FRAMEWORK IN THE ENTERPRISE ARCHITECTURE OF SECONDARY SCHOOLS Lubis, Rivaldi; Panjaitan, Erwin Setiawan
JURTEKSI (jurnal Teknologi dan Sistem Informasi) Vol. 10 No. 3 (2024): Juni 2024
Publisher : Lembaga Penelitian dan Pengabdian Kepada Masyarakat (LPPM) STMIK Royal Kisaran

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33330/jurteksi.v10i3.3235

Abstract: The widespread use of information and communication technology (ICT) has enhanced effectiveness and quality in management, research, and education at educational institutions. In this digital age, secondary schools are required to improve operational efficiency and educational strategies through the use of information technology. Therefore, the implementation of an enterprise architecture (EA) framework is crucial to ensure a strategic alignment between educational goals and technology. However, before implementing the framework, schools must evaluate various factors that influence the suitability and effectiveness of EA, including current technology needs, staff competencies, existing infrastructure conditions, and other factors. This study gathers and analyzes data from related studies, and the results indicate the importance of understanding EA principles to optimize academic and administrative processes. By considering variables in the selection of the EA framework, evaluating school readiness, and identifying existing challenges, this research aims to assist secondary schools in effectively implementing EA. The expected outcome of this research provides theoretical support for the adoption of EA, thus facilitating more efficient strategic and operational planning in secondary schools. Keywords: enterprise architecture; framework; information and communication technology; secondary school. Abstrak: Penggunaan teknologi informasi dan komunikasi (TIK) secara luas telah meningkatkan efektivitas dan kualitas dalam manajemen, penelitian, dan pendidikan di institusi pendidikan. Di era digital ini, sekolah menengah dituntut untuk meningkatkan efisiensi operasional dan strategi pendidikan melalui pemanfaatan teknologi informasi. Oleh karena itu, penerapan framework arsitektur enterprise (EA) menjadi penting untuk memastikan aliansi strategis antara tujuan pendidikan dan teknologi. Namun, sebelum penerapan framework dilakukan, sekolah harus mengevaluasi berbagai faktor yang mempengaruhi kesesuaian dan efektivitas EA, termasuk kebutuhan teknologi terkini, kompetensi staf, kondisi infrastruktur yang ada, dan faktor lainnya. Penelitian ini mengumpulkan dan menganalisis data dari studi terkait, hasilnya menunjukkan bahwa pentingnya pemahaman tentang prinsip-prinsip EA untuk mengoptimalkan proses akademik dan administratif. Dengan mempertimbangkan variabel-variabel dalam pemilihan framework EA, evaluasi kesiapan sekolah, dan identifikasi tantangan yang ada, penelitian ini bertujuan untuk membantu sekolah menengah dalam mengimplementasikan EA secara efektif. Diharapkan hasil dari penelitian ini memberikan kontribusi teoretis yang mendukung pengadopsian EA, sehingga memfasilitasi perencanaan strategis dan operasional yang lebih efisien di sekolah menengah.Kata kunci:arsitektur enterprise; framework; sekolah menengah; teknologi informasi dan komunikasi.

BERT Model Implementation for Dynamic Sentiment Analysis of Pertamina on Social Media X Ronsen Purba; Rivaldi Lubis; Nadya Sikana; Gilbert Fernando Situmorang
Engineering Science Letter Vol. 4 No. 02 (2025): Engineering Science Letter
Publisher : The Indonesian Institute of Science and Technology Research

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56741/IISTR.esl.001139

This study aims to investigate the dynamics of public sentiment on platform X in response to the Pertamina corruption scandal, exploring how trust and perception shifted before and after the incident. Utilizing BERT-based sentiment classification model trained on real-world social media posts, the model achieved a validation loss of 0.5078 and an F1-score of 82.12%, demonstrating strong predictive performance for large-scale sentiment analysis. Results revealed a significant rise in negative sentiment and a decline in positive sentiment following the public disclosure of the scandal on February 25, 2025, reflecting a deep erosion of public trust in Pertamina. Qualitative thematic analysis further identified a shift from neutral or positive discussions focused on service quality and innovation to emotionally charged critiques emphasizing betrayal, distrust and institutional failure. These findings highlight the value of integrating deep learning classification with qualitative insights to monitor real-time public opinion and institutional reputation. The study underscores the critical need for transparency and effective communication strategies during reputational crises to rebuild public confidence. Limitations include the focus on a single social media platform, suggesting future research should incorporate cross-platform and multilingual analyses. Practically, this research offers actionable insights for corporate crisis management and contributes to understanding social media’s role in shaping public trust and accountability in the digital age.

Hybrid Machine Learning for Crime Prediction in Indonesia toward Society 5.0 Nadya Sikana; Rivaldi Lubis; Gilbert Fernando Situmorang; Naomi Prisella
Engineering Science Letter Vol. 4 No. 03 (2025): Engineering Science Letter
Publisher : The Indonesian Institute of Science and Technology Research

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.56741/IISTR.esl.001359

Crime remains a major social challenge in Indonesia, requiring innovative approaches to enhance prevention and law enforcement. This study proposes a hybrid machine learning framework that integrates the Temporal Fusion Transformer (TFT) for time-series forecasting and Extreme Gradient Boosting (XGBoost) for classification and feature analysis. Using socio-economic and demographic data from the Indonesian Central Bureau of Statistics (2010-2023) across 38 provinces, the framework aims to predict crime incidence and classify crime resolution effectiveness. The results show that TFT effectively captures temporal dependencies, achieving robust forecasting accuracy (R2 = 0.9893), while XGBoost delivers high classification performance (Accuracy = 98.87%). Feature importance analysis highlights the dominant role of case resolution rate, government consumption expenditure, school participation rates and life expectancy in shaping crime patterns. Compared to baseline models such as LSTM and Random Forest, the hybrid TFT + XGBoost approach demonstrates superior balance between accuracy, robustness and interpretability. These findings provide actionable insights for policymakers to design data-driven crime prevention strategies, align with Indonesia’s digital transformation agenda, and support the vision of Society 5.0.

Enhancing students’ digital competencies through basic web training at SMKS Indonesia Membangun 1 Gilbert Fernando Situmorang; Sunaryo Winardi; Rivaldi Lubis; Nadya Sikana; Ronsen Purba
Transformasi: Jurnal Pengabdian Masyarakat Vol. 21 No. 2 (2025): Transformasi Desember
Publisher : LP2M Universitas Islam Negeri Mataram

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20414/transformasi.v21i2.14414

[Bahasa]: Keterbatasan siswa SMKS Indonesia Membangun 1, khususnya program keahlian Teknik Komputer dan Jaringan (TKJ), dalam mengonversi pengetahuan pemrograman dasar menjadi artefak digital yang aplikatif menjadi latar belakang kegiatan ini. Pembelajaran yang berfokus pada sistem jaringan menyebabkan penerapan konsep pemrograman belum terintegrasi dengan kebutuhan pengembangan digital. Kegiatan pengabdian masyarakat ini bertujuan untuk menjembatani kesenjangan tersebut melalui pelatihan pengembangan antarmuka web berbasis HTML dan CSS. Program dirancang dengan pendekatan Participatory Action Research (PAR) yang mencakup analisis kebutuhan, perancangan pembelajaran, pelaksanaan pelatihan tatap muka, dan evaluasi. Sebanyak 55 siswa berpartisipasi dalam program. Evaluasi kuantitatif dilakukan melalui pre-test dan post-test menggunakan 10 soal objektif pada platform Kahoot!, dengan analisis data secara deskriptif berdasarkan skor total dan jumlah jawaban yang benar. Skor Kahoot! merepresentasikan akurasi dan kecepatan jawaban. Capaian praktik dievaluasi melalui produk antarmuka web sederhana yang dikembangkan oleh siswa dalam siklus PAR. Hasil menunjukkan peningkatan kompetensi yang signifikan. Rerata skor peserta meningkat dari 2.041,91 menjadi 4.828,78, dan rerata jawaban benar meningkat dari 2,91 menjadi 5,98 dari 10 soal. Meski efektif, evaluasi berbasis daring menghadapi kendala teknis seperti ketergantungan pada koneksi internet. Oleh karena itu, disarankan agar kegiatan serupa di masa depan mengembangkan instrumen penilaian yang kompatibel dengan pembelajaran luring serta mempertimbangkan pendekatan low-code/no-code sebagai strategi pedagogis inklusif untuk memfasilitasi beragam tingkat kesiapan siswa dalam menghasilkan artefak digital. Kata Kunci: desain web, keterampilan digital, kurikulum independen, teknologi [English]: The limitations of SMKS Indonesia Membangun 1 students, particularly those in the Computer and Network Engineering (TKJ) program, in converting basic programming knowledge into applicable digital artifacts served as the background for this activity. Learning focused on network systems has resulted in the application of programming concepts not being integrated with digital development needs. This community service activity aims to bridge this gap by providing training in developing HTML- and CSS-based web interfaces. The program was designed using a Participatory Action Research (PAR) approach that included needs analysis, learning design, face-to-face training implementation, and evaluation. A total of 55 students participated in the program. Quantitative evaluation was conducted through pre-tests and post-tests using 10 objective questions on the Kahoot! platform, with descriptive data analysis based on the total score and the number of correct answers. The Kahoot! score represents the accuracy and speed of answers. Practical achievements were evaluated through a simple web interface product developed by students in the PAR cycle. The results showed a significant increase in competency. The average score of participants increased from 2,041.91 to 4,828.78, and the average correct answer rate increased from 2.91 to 5.98 out of 10 questions. Although effective, online-based evaluations face technical challenges such as dependence on an internet connection. Therefore, it is recommended that similar activities in the future develop assessment instruments that are compatible with offline learning and consider low-code/no-code approaches as inclusive pedagogical strategies to facilitate diverse levels of student readiness in producing digital artifacts. Keywords: web design, digital skills, independent curriculum, technology

OPTIMIZING CYBER ATTACK SIMULATION AS A RESPONSE TO ESCALATING SECURITY THREATS USING A MACHINE LEARNING APPROACH Rivaldi Lubis; Apriyanto Halim; Felix Jansen Tanjaya; Tandri
JURTEKSI (jurnal Teknologi dan Sistem Informasi) Vol. 12 No. 3 (2026): Juni 2026
Publisher : Lembaga Penelitian dan Pengabdian Kepada Masyarakat (LPPM) STMIK Royal Kisaran

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33330/jurteksi.v12i3.4411

Abstract: The growing intensity of cyber attacks, marked by rapid, large-scale, automated, and adaptive execution, requires analytical methods that represent the diversity of network environments, including variations in target platforms such as IoT, traditional networks, and hybrid infrastructures. This study compares machine learning models for cyber attack classification under heterogeneous environmental conditions and formulates a conceptual optimization framework based on model performance. Four publicly available benchmark datasets were used, namely UNB CIC IoT 2023, UNB CIC IDS-2018, UNSW-NB15, and a Kaggle cyber security attacks dataset, comprising approximately 40,000 to over 3.6 million records and 25 to 80 features across IoT, conventional, and mixed network environments. Random Forest, XGBoost, Multilayer Perceptron, and Transformer were implemented within a unified pipeline involving preprocessing, feature selection, and Bayesian Optimization-based hyperparameter tuning. All models achieved F1-score and Cohen's Kappa above 96%, with XGBoost performing best (97.80%, 97.26%), followed by Random Forest (97.78%, 96.96%) and Transformer (97.44%, 96.82%), while MLP scored lowest (96.74%, 96.00%), a gap below one percentage point. Confusion matrix analysis revealed persistent misclassification in minority and overlapping attack classes, informing a proposed adaptive cyber attack simulation optimization framework. Keywords: cyber attacks; optimization; machine learning; environmental variability. Abstrak: Meningkatnya intensitas serangan siber yang berlangsung cepat, masif, otomatis, dan adaptif menuntut pendekatan analitis yang merepresentasikan keragaman lingkungan jaringan, termasuk perbedaan karakteristik platform sasaran seperti Internet of Things (IoT), jaringan konvensional, dan infrastruktur hibrida. Penelitian ini membandingkan model machine learning untuk klasifikasi serangan siber pada kondisi lingkungan heterogen, sekaligus menyusun kerangka optimasi konseptual berdasarkan performa model. Empat dataset benchmark publik digunakan, yaitu UNB CIC IoT 2023, UNB CIC IDS-2018, UNSW-NB15, serta dataset Kaggle cyber security attacks, dengan jumlah data berkisar 40.000 hingga lebih dari 3,6 juta rekaman dan 25 sampai 80 fitur, mewakili lingkungan IoT, konvensional, dan campuran. Random Forest, XGBoost, Multilayer Perceptron, dan Transformer diimplementasikan melalui pipeline terpadu mencakup pra-pemrosesan, seleksi fitur, dan optimasi hyperparameter berbasis Bayesian Optimization. Seluruh model mencapai F1-score dan Cohen's Kappa di atas 96%, dengan XGBoost menunjukkan performa terbaik (97,80%, 97,26%), diikuti Random Forest (97,78%, 96,96%) dan Transformer (97,44%, 96,82%), sementara MLP mencatat skor terendah (96,74%, 96,00%), dengan selisih kurang dari satu poin persentase. Analisis confusion matrix mengungkap misklasifikasi yang konsisten pada kelas minoritas dan serangan dengan karakteristik serupa, yang menjadi dasar kerangka optimasi simulasi serangan siber adaptif yang diusulkan. Kata kunci: serangan siber; optimasi; machine learning; variabilitas lingkungan

Klasifikasi Perilaku Keuangan UMKM Medan dengan Machine Learning dan SHAP Saliman Saliman; Rivaldi Lubis; Sunaryo Winardi; Mustika Ulina
BIOS : Jurnal Teknologi Informasi dan Rekayasa Komputer Vol 7 No 2 (2026): September (In Progress)
Publisher : Puslitbang Sinergis Asa Professional

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37148/bios.v7i2.224

Good financial behavior is an important factor in sustaining micro, small, and medium enterprises (MSMEs), particularly in financial recording, debt management, and investment planning. Previous research using Structural Equation Modeling Partial Least Squares (SEM-PLS) identified financial attitude as the dominant factor influencing MSMEs’ financial behavior in Medan City. Based on these findings, this study develops a complementary machine learning-based approach to classify MSMEs’ financial behavior at the individual level and evaluate its consistency with SEM results through SHAP-based Explainable Artificial Intelligence (XAI). The dataset consists of 100 MSME respondents with seven main features, including three financial constructs and four demographic variables. Three ensemble algorithms, namely CatBoost, XGBoost, and Random Forest, were evaluated using hold-out and Stratified 5-Fold Cross-Validation. The results show that CatBoost achieved the best performance with 80.00% accuracy and 82.46% F1-score. SHAP analysis confirmed the dominance of attitude score and revealed the significant predictive contribution of demographic variables. Integrating machine learning and SHAP is an effective complementary approach to extend the understanding of MSMEs’ financial behavior comprehensively.

PERBANDINGAN KESESUAIAN PENDAPAT MANUSIA DAN AI (CHATGPT, GEMINI, DEEPSEEK) MENGGUNAKAN PENDEKATAN GROUND TRUTH Kelvin Kelvin; Sunaryo Winardi; Handoko Handoko; Erwin Setiawan Panjaitan; Rivaldi Lubis
JOURNAL OF SCIENCE AND SOCIAL RESEARCH Vol. 9 No. 3 (2026): June 2026
Publisher : Smart Education

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.54314/jssr.v9i3.6784

Abstract: The rapid advancement of Large Language Models (LLMs) has significantly improved sentiment analysis capabilities. However, the extent to which these models produce sentiment classifications consistent with human judgment remains an important research question. This study aims to evaluate and compare the agreement of ChatGPT, Gemini, and DeepSeek with human-generated ground truth in sentiment analysis. A total of 1,497 product reviews were collected from the Sephora Products and Skincare Reviews dataset. Three independent annotators labeled each review as positive, neutral, or negative to establish the ground truth. The annotation reliability achieved a Fleiss' Kappa coefficient of 0.7053, indicating substantial agreement and confirming the reliability of the ground truth for evaluation purposes. Subsequently, the three LLMs performed sentiment classification using an Expectation–Role–Action (ERA) prompting strategy. Model performance was assessed using Cohen's Kappa to measure agreement with the ground truth and Spearman's rank correlation to evaluate the consistency of sentiment rankings. The results show that DeepSeek achieved the highest performance, with an average Cohen's Kappa of 0.753 and an average Spearman's correlation coefficient of 0.860, followed by Gemini (κ = 0.662; ρ = 0.834), while ChatGPT demonstrated the lowest agreement (κ = 0.223; ρ = 0.367). These findings indicate that the three LLMs exhibit significantly different levels of agreement with human judgment, with DeepSeek producing sentiment classifications that most closely align with the established ground truth. Keywords: Large Language Models, Sentiment Analysis, Ground Truth, Cohen's Kappa, Spearman's Rank Correlation. Abstrak: Perkembangan Large Language Models (LLMs) telah meningkatkan kemampuan analisis sentimen berbasis kecerdasan buatan. Namun, tingkat kesesuaian hasil klasifikasi sentimen yang dihasilkan oleh berbagai LLM terhadap penilaian manusia masih memerlukan evaluasi yang komprehensif. Penelitian ini bertujuan membandingkan tingkat kesesuaian hasil analisis sentimen ChatGPT, Gemini, dan DeepSeek terhadap ground truth yang diperoleh melalui anotasi manusia. Penelitian menggunakan 1.497 ulasan produk dari dataset Sephora Products and Skincare Reviews. Sebanyak tiga anotator independen melakukan pelabelan sentimen ke dalam kategori positif, netral, dan negatif untuk membentuk ground truth. Hasil pengujian reliabilitas menunjukkan nilai Fleiss' Kappa sebesar 0,7053, yang mengindikasikan tingkat kesepakatan Substantial Agreement, sehingga ground truth layak digunakan sebagai acuan evaluasi. Selanjutnya, ketiga model LLM melakukan klasifikasi sentimen menggunakan prompt berbasis Expectation–Role–Action (ERA). Tingkat kesesuaian hasil klasifikasi dievaluasi menggunakan Cohen's Kappa, sedangkan konsistensi hubungan dengan ground truth dianalisis menggunakan korelasi Spearman. Hasil penelitian menunjukkan bahwa DeepSeek memberikan performa terbaik dengan rata-rata Cohen's Kappa sebesar 0,753 dan rata-rata korelasi Spearman sebesar 0,860, diikuti oleh Gemini (κ = 0,662; ρ = 0,834), sedangkan ChatGPT memperoleh tingkat kesesuaian terendah (κ = 0,223; ρ = 0,367). Temuan ini menunjukkan bahwa terdapat perbedaan kemampuan interpretasi sentimen antar model LLM, dengan DeepSeek menghasilkan klasifikasi yang paling mendekati penilaian manusia pada dataset yang digunakan. Kata kunci: Large Language Model, Analisis Sentimen, Ground Truth, Cohen's Kappa, Spearman Correlation.

Canva-based digital portfolio website training as a community service intervention to enhance students' digital literacy Rivaldi Lubis; Ng Poi Wong; Sunario Megawan; Tanti; Siti Endang Kristiani Gea
Transformasi: Jurnal Pengabdian Masyarakat Vol. 22 No. 1 (2026): Transformasi Juni
Publisher : LP2M Universitas Islam Negeri Mataram

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20414/transformasi.v22i1.15273

[Bahasa]: Literasi digital merupakan kompetensi penting bagi siswa sekolah menengah dalam menghadapi kebutuhan pendidikan lanjutan dan dunia kerja yang semakin terdigitalisasi. Salah satu bentuk literasi digital yang relevan adalah kemampuan menyusun portofolio digital berbasis website sebagai sarana representasi identitas, kompetensi, dan capaian personal. Namun, pembelajaran pembuatan website di sekolah masih didominasi pendekatan teknis berbasis pemrograman yang relatif sulit diakses oleh pemula dan belum berorientasi pada luaran portofolio yang siap digunakan. Program pengabdian kepada masyarakat ini bertujuan meningkatkan pemahaman siswa tentang portofolio digital dan kemampuan menghasilkan website portofolio menggunakan Canva Website Builder. Program menerapkan pendekatan pelatihan partisipatif dan berorientasi tindakan melalui pelatihan, praktik terbimbing, pengembangan portofolio mandiri, dan evaluasi selama dua hari. Kegiatan dilaksanakan di SMA Swasta W.R. Supratman 1 Medan dengan melibatkan 32 siswa. Hasil evaluasi menunjukkan peningkatan skor pemahaman konseptual yang signifikan secara statistik dari rata-rata 4,03 menjadi 6,81 (skala 0–10; n=32; uji-t berpasangan, p<0,001; d=1,64), dengan 28 peserta (87,5%) menunjukkan peningkatan skor individual. Sebanyak 30 dari 32 peserta (93,8%) berhasil menerbitkan website portofolio yang dapat diakses dan memenuhi kriteria minimum rubrik penilaian, sementara 30 peserta yang mengisi kuesioner persepsi (n=30) melaporkan penilaian positif terhadap kualitas dan relevansi pelatihan. Temuan ini menunjukkan bahwa pendekatan pelatihan berbasis platform non-koding layak diterapkan dan diterima secara positif dalam mendukung penguatan literasi digital di tingkat pendidikan menengah, serta membuka peluang bagi transformasi pedagogi digital yang lebih luas dan berkelanjutan pada jenjang sekolah menengah di masa mendatang. Kata Kunci: keterampilan digital, Canva, desain web, portofolio digital [English]: Digital literacy is a critical competency for secondary school students in responding to the increasing demands of higher education and digitally driven labor markets. One relevant form of digital literacy is the ability to develop a website-based portfolio that represents personal identity, competencies, and achievements. However, website development learning in schools is dominated by programming-oriented approaches that are less accessible to beginners and insufficiently focused on producing ready-to-use portfolios. This program aims to improve students’ understanding of digital portfolios and their ability to create publishable portfolio websites using Canva Website Builder. It employed a participatory, action-oriented training approach involving instruction, guided practice, independent portfolio development, and evaluation over two days. The program was conducted at SMA Swasta W.R. Supratman 1 Medan and involved 32 students. Evaluation results indicated a statistically significant increase in conceptual understanding scores, from a mean of 4.03 to 6.81 (0–10 scale; n = 32; paired t-test, p < .001; d = 1.64), with 28 participants (87.5%) showing an individual-level score increase. Thirty of the 32 participants (93.8%) successfully published an accessible portfolio website meeting the minimum rubric criteria, while 30 participants who completed the perception questionnaire (n = 30) reported positive evaluations of training quality and relevance. These findings suggest that non-coding, platform-based training is a feasible and well-received approach for strengthening digital literacy at the secondary education level, and point toward broader potential for sustained digital pedagogy transformation in secondary schools going forward. Keywords: digital skills, Canva, web design, digital portfolio

An Intelligent System for Detecting Online Gambling Promotion on Social Media Based on BERT and Adaptive Browser Extension Gilbert Fernando Situmorang; Juliana Damayanti Manurung; Rivaldi Lubis; Nadya Sikana; Kenneth Lionggo
Jurnal Nasional Pendidikan Teknik Informatika: JANAPATI Vol. 14 No. 3 (2025)
Publisher : Prodi Pendidikan Teknik Informatika Universitas Pendidikan Ganesha

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23887/janapati.v14i3.103511

The proliferation of online gambling promotions on social media has created serious social and legal problems in Indonesia, yet existing detection approaches are largely post-hoc, rely on general purpose language models, and struggle to identify implicit and domain specific promotional language that is deliberately used to evade moderation. Moreover, most prior studies focus solely on offline model evaluation and are rarely integrated into real time, deployable systems that can directly protect users. To address these limitations, this study proposes an intelligent detection system utilizing IndoBERT enhanced with Domain Adaptive Pre-Training on a corpus of Indonesian online gambling related text, followed by fine-tuning for text classification. The adapted model demonstrates superior performance, achieving an F1-Score of 98.58% and outperforming baseline approaches including TF-IDF+SVM, BiLSTM, multilingual BERT, and IndoBERT without domain adaptation. To bridge the gap between model development and real-world application, the proposed model is further integrated into an adaptive browser extension capable of scanning, classifying, and filtering social media content in real-time. Functional testing on YouTube and X shows that the system effectively detects and masks online gambling promotional content without disrupting neutral or anti-gambling discourse. This research contributes both methodologically, by demonstrating the effectiveness of domain adaptive pre-training for detecting implicit promotional language, and practically, by delivering a deployable system that provides proactive protection for social media users.

Title

Found 10 Documents
Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title Search

Found 10 Documents Search

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Title

Found 10 Documents
Search