Jurnal Teknik Informatika UNIKA Santo Thomas

Evaluasi Komparatif Naive Bayes dan SVM pada Sentimen Berbahasa Indonesia

2026-07-21T06:13:51+00:00

Purpose: This study compares Multinomial Naive Bayes and Support Vector Machine for three-class sentiment classification of Indonesian text. Design/methods/approach: A controlled synthetic dataset of 3,000 online-learning comments was generated with balanced positive, neutral, and negative labels. The data include formal and informal expressions, negation, mixed sentiment, and spelling variation. Preprocessing consisted of case folding, cleaning, normalization, stopword removal, and negation handling. Unigram and bigram features were weighted using TF-IDF. Models were tuned through five-fold cross-validation and evaluated on a stratified 20% test set using accuracy, macro precision, macro recall, macro F1-score, and confusion matrices. Findings/results: Naive Bayes achieved 82.83% accuracy and 82.82% macro F1, whereas SVM achieved 82.83% accuracy and 82.82% macro F1. The error analysis indicates that mixed-polarity sentences, negation, and neutral expressions are the most challenging patterns. Conclusions: Both models provide equivalent classification performance in this controlled experiment, although their modeling assumptions differ. Because the dataset is synthetic, the findings should be validated using real-world data.

Sistem Keamanan Toko Cerdas Berbasis IoT Dengan Pemantauan Visual Dan Notifikasi Real-Time

2026-04-12T18:23:37+00:00

Penelitian ini bertujuan untuk menciptakan dan menilai sistem keamanan toko cerdas berbasis Internet of Things (IoT) yang dapat memberikan pemantauan visual dan notifikasi real-time kepada pemilik bisnis. Sistem ini dirancang menggunakan ESP32-CAM, sensor PIR untuk mendeteksi pergerakan manusia, sensor ultrasonik HC-SR04 untuk memvalidasi jarak objek, dan modul RTC untuk pengaturan jadwal aktif otomatis berdasarkan jam operasional toko dan jam istirahat, dengan notifikasi dan bukti visual yang dikirim melalui aplikasi Telegram. Metode yang digunakan adalah pendekatan kuantitatif berdasarkan model Penelitian dan Pengembangan (R&D), dengan tahapan termasuk analisis kebutuhan, desain, pembuatan prototipe, dan pengujian dan lapangan. Akurasi deteksi sensor, waktu reaksi pengiriman notifikasi, tingkat keberhasilan sistem, dan jumlah alarm palsu termasuk di antara karakteristik yang akan diteliti. Data dikumpulkan melalui observasi terorganisir, perekaman log sistem, dan wawancara pengguna. Hasil pengujian menunjukkan bahwa sistem dapat mendeteksi pergerakan secara optimal pada jarak 1-3 meter, mengukur jarak objek secara tepat pada rentang 10-30 cm, dan mengirimkan pesan dengan cepat dan andal dengan tingkat keberhasilan yang tinggi. Integrasi dua sensor telah terbukti meningkatkan akurasi dan mengurangi false alarm, sementara modul RTC memastikan bahwa sistem beroperasi sesuai jadwal, menghasilkan solusi yang dianggap efektif, responsif, terjangkau, dan cocok untuk diimplementasikan di toko-toko kecil dan menengah.

Model Budidaya Tanaman Adaptif Berbasis Digital Twin dan Kecerdasan Buatan untuk Pertanian Presisi

2026-07-09T05:17:58+00:00

Precision agriculture requires a system capable of integrating agronomic data, simulations, and artificial intelligence to support adaptive crop management decision-making. However, most previous research has focused on monitoring or simulation without combining predictive and recommendation functions within a single integrated framework. This study aims to develop an Adaptive Crop Cultivation Model Based on Digital Twins and Artificial Intelligence for Precision Agriculture using the Design Science Research (DSR) approach. The proposed model integrates crop cultivation data from the Crop Recommendation Dataset (2,200 data points) and soil characteristic data from SoilGrids (5,000 data points) through the stages of data acquisition, data cleaning, normalization, and feature engineering. The developed framework consists of a Digital Twin Layer, an AI and Simulation Layer, and an Adaptive Recommendation Layer to support real-time simulation, prediction, and crop cultivation recommendations. The results demonstrate that the integration of Digital Twin and artificial intelligence can create a virtual representation of crops, support predictions of water requirements, fertilization needs, and disease risks, and generate adaptive cultivation recommendations based on agronomic conditions. The proposed model has the potential to improve resource management efficiency, enhance the quality of decision-making, and support the implementation of smarter and more sustainable precision agriculture.

RANCANG BANGUN SISTEM PEMANTAUAN DAN PENCEGAHAN KEBAKARAN BERBASIS IOT PADA TPDK PEDURUNGAN

2026-03-16T05:58:44+00:00

Kebakaran merupakan ancaman serius yang dapat mengakibatkan kerugian materiil dan korban jiwa yang signifikan, namun di TPDK Pedurungan pengawasan gedung masih bersifat konvensional dan bergantung pada kehadiran fisik petugas. Penelitian ini bertujuan merancang sistem pemantauan dan pencegahan kebakaran berbasis Internet of Things (IoT) menggunakan metode Research and Development (R&D) untuk meningkatkan efisiensi deteksi dini. Sistem ini mengintegrasikan mikrokontroler ESP32 dengan sensor DHT11 untuk memantau suhu serta flame sensor untuk mendeteksi api, didukung indikator visual LED merah-hijau dan buzzer sebagai alarm lokal. Seluruh data sensor disimpan secara otomatis dalam database Firebase dan informasi kondisi ruangan dapat diakses secara real-time melalui aplikasi Telegram sebagai media notifikasi otomatis. Hasil pengujian menunjukkan bahwa sistem mampu mendeteksi indikasi api dan kenaikan suhu ekstrem dengan akurasi tinggi serta respon transmisi data yang cepat. Implementasi sistem ini memberikan solusi preventif yang efisien bagi TPDK Pedurungan dalam meminimalisir risiko kebakaran serta melindungi keamanan aset dokumen kependudukan secara optimal.

Analisis Sentimen Publik terhadap Program Makan Bergizi Gratis Menggunakan Support Vector Machine: Komparasi Teknik Ekstraksi Fitur pada Data Komentar YouTube

2026-07-31T09:23:33+00:00

This study aims to analyze public sentiment toward the Free Nutritious Meal Program (MBG) based on opinions expressed on social media. The research data were collected through a web scraping process using relevant keywords, resulting in a total of 400 comments. After undergoing a series of text preprocessing stages, including case folding, normalization, text cleaning, stopword removal, stemming, and tokenization, 242 comments were retained as the final dataset for analysis. Sentiment labeling was performed automatically using the Indonesia Sentiment Lexicon (InSet), and the comments were categorized into three sentiment classes: positive, negative, and neutral. Sentiment classification was conducted using the Support Vector Machine (SVM) algorithm with three feature extraction techniques, namely Term Presence, Bag of Words (BoW), and Term Frequency–Inverse Document Frequency (TF-IDF). The model performance was evaluated using a confusion matrix and 5-fold cross-validation, combined with the Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance. The results indicate that neutral sentiment predominated, accounting for 52.89% of the total comments, followed by positive sentiment (31.82%) and negative sentiment (15.29%). These findings suggest that the majority of the public adopted a moderate stance by providing evaluations, recommendations, and expectations regarding the implementation of the Free Nutritious Meal Program (MBG).