Article Per Year (5 Year)
p-Index From 2021 - 2026
4.281
P-Index
This Author published in this journals
All Journal Jurnal Sistem dan Informatika Indonesian Journal of Electrical Engineering and Computer Science Journal La Multiapp Edutik : Jurnal Pendidikan Teknologi Informasi dan Komunikasi Jurnal Minfo Polgan (JMP) Abdimas : Jurnal Pengabdian Kepada Masyarakat Innovative: Journal Of Social Science Research Journal of Embedded Systems, Security and Intelligent Systems Edutik : Jurnal Pendidikan Teknologi Informasi dan Komunikasi Journal of Vocational, Informatics and Computer Education
Irene Realyta Halldy Trosi Tangkawarow
Informatics Department, Engineering Faculty, Universitas Negeri Manado
Aerospace Engineering Humanities Automotive Engineering Chemical Engineering, Chemistry & Bioengineering Civil Engineering, Building, Construction & Architecture Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Education Electrical & Electronics Engineering Engineering Environmental Science Languange, Linguistic, Communication & Media Law, Crime, Criminology & Criminal Justice Library & Information Science Materials Science & Nanotechnology Mathematics Public Health Social Sciences Other

Published : 34 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 34 Documents
Search

 1   2   3   4 
Pemanfaatan Algoritma Convolutional Neural Networks (CNN) untuk Analisis Opini Publik terhadap Pemilihan Umum melalui Media Sosial X Claudia Pither; Irene Realyta Haldy Trosi Tangkawarow; Audy Aldrin Kenap
Edutik : Jurnal Pendidikan Teknologi Informasi dan Komunikasi Vol. 6 No. 4 (2026): EduTIK : Agustus 2026
Publisher : Jurusan PTIK Universitas Negeri Manado

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.67142/edutik.v6i4.523

Abstract

ABSTRAK  Studi ini bertujuan untuk mengklasifikasikan sentimen publik terkait Pemilihan Suara Ulang (PSU) guna memetakan opini publik digital di media sosial X yang sangat penting bagi peneliti dan administrator pemilu dalam memahami respons masyarakat terhadap sengketa pemilu dengan kondisi data yang sangat tidak seimbang. Sebanyak 1.016 tweet dikumpulkan dan diproses melalui tahapan pra-pemrosesan teks meliputi pembersihan, normalisasi, dan stemming, yang kemudian dilabeli menggunakan leksikon kontekstual diperluas serta titik data manual. Untuk mengatasi ketidakseimbangan kelas ekstrem tanpa menyebabkan kebocoran data, algoritma Random Over-Sampling (ROS) diterapkan secara eksklusif pada set pelatihan sebelum melatih model Convolutional Neural Network (CNN) berbasis embedding Word2Vec (100 dimensi, ukuran jendela 5, 15 epoch, ukuran batch 32) untuk dibandingkan dengan baseline Naive Bayes + TF-IDF + ROS. Berdasarkan distribusi dataset yang terdiri dari 79,4% Netral, 18,3% Positif, dan 2,3% Negatif, hasil eksperimen menunjukkan bahwa model CNN + Word2Vec + ROS yang diusulkan berhasil mencapai akurasi validasi sebesar 92,20% dan skor F1 Makro sebesar 0,78, terbukti mengungguli model dasar Naive Bayes yang hanya mencapai akurasi 80,49% dan skor F1 Makro 0,73. Meskipun integrasi Word2Vec dan ROS berhasil mengurangi keterbatasan pengklasifikasi berbasis frekuensi dalam wacana politik yang timpang, skor F1 untuk kelas negatif minoritas yang bernilai 0,55 menyoroti bahwa mendeteksi opini kritis masih menjadi tantangan yang memerlukan penyempurnaan metodologis lebih lanjut. Secara keseluruhan, studi ini memberikan kontribusi penting pada analisis sentimen politik di Indonesia dengan mengevaluasi potensi pembelajaran mendalam (deep learning) dalam menyediakan kerangka kerja pemantauan sentimen elektoral yang kuat, sekaligus menekankan pentingnya ketelitian metodologis dalam menyeimbangkan deteksi sudut pandang minoritas secara cermat. ABSTRACT  This study aims to classify public sentiment related to the Re-vote Election (PSU) in order to map digital public opinion on social media X which is very important for researchers and election administrators in understanding public responses to election disputes with highly imbalanced data conditions. A total of 1,016 tweets were collected and processed through text pre-processing stages including cleaning, normalization, and stemming, which were then labeled using an expanded contextual lexicon as well as manual data points. To address extreme class imbalance without causing data leakage, the Random Over-Sampling (ROS) algorithm was applied exclusively to the training set before training a Word2Vec embedding-based Convolutional Neural Network (CNN) model (100 dimensions, window size 5, 15 epochs, batch size 32) to be compared with the baseline Naive Bayes + TF-IDF + ROS. Based on the dataset distribution consisting of 79.4% Neutral, 18.3% Positive, and 2.3% Negative, the experimental results show that the proposed CNN + Word2Vec + ROS model successfully achieved a validation accuracy of 92.20% and a Macro F1 score of 0.78, proven to outperform the baseline Naive Bayes model which only achieved an accuracy of 80.49% and a Macro F1 score of 0.73. Although the integration of Word2Vec and ROS successfully mitigated the limitations of frequency-based classifiers in imbalanced political discourse, the F1 score for the minority negative class of 0.55 highlights that detecting critical opinions remains a challenge that requires further methodological refinement. Overall, this study makes an important contribution to political sentiment analysis in Indonesia by evaluating the potential of deep learning in providing a robust electoral sentiment monitoring framework, while emphasizing the importance of methodological rigor in carefully balancing minority viewpoint detection.
Analisis Sentimen Pelanggaran Politik Uang pada Pilkada Menggunakan Integrasi SMOTE dan Long Short-Term Memory (LSTM) Keiza Sasako; Irene Realyta Haldy Trosi Tangkawarow; Efraim Ronald S Moningkey
Edutik : Jurnal Pendidikan Teknologi Informasi dan Komunikasi Vol. 6 No. 4 (2026): EduTIK : Agustus 2026
Publisher : Jurusan PTIK Universitas Negeri Manado

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.67142/edutik.v6i4.524

Abstract

ABSTRAK               Praktik politik uang merupakan ancaman serius terhadap integritas Pemilihan Kepala Daerah (Pilkada) di Indonesia, di mana opini dan keluhan publik banyak disuarakan melalui media sosial X. Namun, analisis sentimen otomatis terhadap isu ini sering terkendala oleh ketimpangan kelas data yang parah antara dokumen teks positif dan negatif. Penelitian ini bertujuan untuk mengimplementasikan alur kerja baru yang mengintegrasikan teknik penyeimbangan data berbasis Synthetic Minority Over-sampling Technique (SMOTE) dan arsitektur sekuensial Long Short-Term Memory (LSTM) untuk mengklasifikasikan sentimen terkait pelanggaran pemilu. Dataset yang digunakan terdiri dari 1.013 unggahan bahasa Indonesia dari platform X yang diekstraksi dari Oktober hingga November 2024. Metode penelitian meliputi prapemrosesan teks (cleaning, case folding, normalisasi, stopword removal, dan tokenisasi), ekstraksi fitur Embedding, aplikasi SMOTE pada ruang vektor, dan klasifikasi menggunakan LSTM 128 memori tersembunyi. Hasil penelitian menunjukkan bahwa penambahan SMOTE berhasil mengatasi masalah bias kelas mayoritas pada model LSTM standar. Model usulan SMOTE-LSTM ini mampu mendongkrak performa deteksi kelas minoritas secara signifikan, menghasilkan nilai presisi sebesar 0,84, recall sebesar 0,83, dan F1-score sebesar 0,83 pada kelas negatif (laporan kecurangan). Kesimpulan dari penelitian ini menunjukkan bahwa integrasi SMOTE dan LSTM memberikan sensitivitas yang kokoh dalam mengenali opini kecurangan pemilu, sehingga sangat layak dipertimbangkan sebagai modul awal sistem pemantauan digital otomatis bagi Badan Pengawas Pemilihan Umum (Bawaslu).  ABSTRACT               The practice of money politics poses a serious threat to the integrity of regional elections (Pilkada) in Indonesia, where public opinions and complaints are widely voiced through the social media platform X. However, automated sentiment analysis on this issue is frequently hindered by a severe data class imbalance between positive and negative text documents. This study aims to implement a new workflow integrating the data balancing technique based on Synthetic Minority Over-sampling Technique (SMOTE) and the sequential architecture of Long Short-Term Memory (LSTM) to classify sentiments regarding election violations. The dataset consists of 1,013 Indonesian posts from platform X, extracted from October to November 2024. The methodology includes text preprocessing (cleaning, case folding, normalization, stopword removal, and tokenization), Embedding feature extraction, the application of SMOTE in the vector space, and classification using an LSTM with 128 hidden memory units. The results indicate that the incorporation of SMOTE successfully resolved the majority class bias inherent in the standard LSTM model. The proposed SMOTE-LSTM model significantly boosted the detection performance of the minority class, yielding a precision of 0.84, a recall of 0.83, and an F1-score of 0.83 for the negative class (fraud reports). The conclusion of this study demonstrates that the integration of SMOTE and LSTM provides robust sensitivity in recognizing election fraud opinions, making it highly viable as an initial automated digital monitoring module for the Elections Supervisory Agency (Bawaslu).
Implementation of the Multinomial Naïve Bayes Algorithm in a Web-Based System for Detecting Online News Hoaxes During the 2024 Elections Juliawati Haribae; Irene R.H.T Tangkawarow; Gladly C. Rorimpandey
Journal of Vocational, Informatics and Computer Education Vol 4, No 3 (2026): September 2026
Publisher : Academic Bright Collaboration

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.66053/voice.v4i3.608

Abstract

Purpose – This study addresses the escalation of disinformation during the 2024 Indonesian General Election by developing an automated hoax detection system. The primary focus is to evaluate the integration of data balancing methods to minimize detection failures in hoax narratives, which often appear less frequently than factual news in real-world scenarios. Methods – The dataset consists of 1,529 unique news documents, comprising 1,020 factual articles from Kompas.com and 509 hoax articles from TurnBackHoax.id. The modelling workflow involves text preprocessing, feature extraction via Term Frequency-Inverse Document Frequency (TF-IDF), and the application of the Synthetic Minority Over-sampling Technique (SMOTE) to address class imbalance. The model was developed using the Multinomial Naive Bayes algorithm and integrated into a web-based platform using the Flask framework. Findings – Results from 5-fold cross-validation demonstrated a stable average accuracy of 92.64% (±1.27%). On an independent test set, the model achieved 89.54% accuracy with a hoax class recall of 0.93. This proves that SMOTE significantly enhances sensitivity in identifying disinformation, reducing the risk of false information bypassing the system. Research implications – The study is limited to textual data from curated sources and cannot yet capture multimedia disinformation or high levels of sarcasm. Future research should explore Transformer-based models for deeper semantic context understanding. Originality – This research provides a functional public verification platform with a response time of under one second. Its novelty lies in a domain-specific detection system that balances algorithmic efficiency with precision on imbalanced real-world election data.
Structural Similarity Analysis of Business Processes in BAWASLU BPM Using the Jaccard Method Nathania Piri; Irene R H T Tangkawarow; Efraim Ronald S Moningkey
Journal of Vocational, Informatics and Computer Education Vol 4, No 3 (2026): September 2026
Publisher : Academic Bright Collaboration

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.66053/voice.v4i3.880

Abstract

Purpose – Business Process Management is essential for standardizing complex procedures in public institutions such as the General Election Supervisory Agency (Bawaslu). However, existing similarity measures, including Graph Edit Distance and semantic approaches, are computationally demanding or depend on extensive execution logs, while their use for auditing public-sector Standard Operating Procedures remains limited. This study evaluates the structural similarity of Bawaslu’s BPM models to identify procedural overlaps and support operational standardization through an efficient method. Methods – A quantitative approach was applied using the Jaccard similarity coefficient. BPMN models from provincial and regency/city levels were represented as directed graphs. Similarity was calculated from the intersection and union of functional nodes, representing activities, and coordination edges, representing transitions. Findings – Across four JDIH processes, the average total structural similarity reached 71.54%. Core activities were highly consistent, with node similarity above 93%, whereas coordination structures varied more substantially, with edge similarity below 67%. The legal-document upload procedure recorded the highest overall similarity score of 0.8181. Research implications – The method can support preliminary identification of redundant procedures, inconsistent coordination patterns, and SOPs requiring deeper review or harmonization. Originality – This study demonstrates the Jaccard coefficient as a lightweight, scalable tool for preliminary structural comparison of public-sector SOPs without intensive computation or extensive process logs.
Co-Authors Adesyan Lahinda Alfiansyah Hasibuan Ameiliani Dianira Barahama Anggreni Audy Aldrin Kenap Biringpasemba, Tiara Christian Yehuda Tendean Christovel Yoshua Tiouw Citra Grace Kalangie Claudia Pither Dandy Pramana Hostiadi Desri Natalis Lasa Dwi Putri Sitanggang Efraim R. S. Moningkey Emor, Stevren Christian Eva Hariyanti Fatonah, Nenden Siti Filemon Blessing Tambuwun Gladly Caren Rorimpandey Gracia I. Ratumbuysang I Made Darma Susila Johanes Mokorimban Pantouw Jovita Pascoal Juliawati Haribae Jullianto Vernando Christofel Rompis Julyeta Paulina Amelia Runtuwene Keiza Sasako Kirei Debora Tasya Palar Kristofel Santa maki, Indriani Fidelia Maki Marthin, Claudia Mohammad Yazdi Pusadan Munir A.J. Tujuwale Murni Kassa Nathania Piri Olgraciella Manitik Prince Cristian Rumate Quido Conferti Kainde Rambega, Sultan Fajri Rompas, Parabelem T. D. Rompas, Parabelem Tino Dolf Semuel Fendy Ngalo Timothy Tendean Yohanes Priyo Atmojo