Article Per Year (5 Year)
p-Index From 2021 - 2026
2.39
P-Index
This Author published in this journals
All Journal Syntax Jurnal Informatika Jurnal Teknologi Informasi dan Ilmu Komputer Jurnal Informatika dan Teknik Elektro Terapan Scientific Journal of Informatics Jurnas Nasional Teknologi dan Sistem Informasi JIKO (Jurnal Informatika dan Komputer) Informatika Pertanian Jurnal TAM (Technology Acceptance Model) IKRA-ITH Informatika : Jurnal Komputer dan Informatika JITTER (Jurnal Ilmiah Teknologi Informasi Terapan) JATI (Jurnal Mahasiswa Teknik Informatika) Tridharmadimas: Jurnal Pengabdian Kepada Masyarakat Jayakarta JUSTIN (Jurnal Sistem dan Teknologi Informasi) Jurnal TAM (Technology Acceptance Model)
Sofi Defiyanti
Universitas Singaperbangsa Karawang
Agriculture, Biological Sciences & Forestry Civil Engineering, Building, Construction & Architecture Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Economics, Econometrics & Finance Electrical & Electronics Engineering Engineering Industrial & Manufacturing Engineering Mathematics Mechanical Engineering Social Sciences

Published : 24 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 24 Documents
Search

 1   2   3 
Analisis Sentimen Teks Code-Mixed Bahasa Indonesia-Jawa Menggunakan Metode Fine-Tuning Model Nusabert Sopian Syauri; Sofi Defiyanti; Dadang Yusup
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Pesatnya pertumbuhan pengguna platform X di Indonesia memicu lahirnya fenomena code-mixingantara Bahasa Indonesia dan Bahasa Jawa dalam komunikasi digital. Teks campuran inimengandung nuansa linguistik lokal yang sulit diproses oleh model NLP berbasis bahasa umum.Penelitian ini mengimplementasikan model NusaBERT dengan teknik fine-tuning untukmelakukan analisis sentimen pada teks code-mixed Indonesia-Jawa yang dikumpulkan dariplatform X. Data sebanyak 1.685 teks diperoleh melalui teknik crawling menggunakan kata kunciberbahasa Jawa, kemudian dilabeli secara otomatis menggunakan AI (Gemini dan Claude) denganvalidasi manual oleh penutur asli. Eksperimen dilakukan dalam sembilan skenario yangmengombinasikan tiga variasi pembagian data (70:30, 80:20, 90:10) dan tiga nilai learning rate(2×10⁻⁵, 3×10⁻⁵, 5×10⁻⁵). Hasil terbaik diperoleh pada Skenario 3 dengan pembagian data 70:30dan learning rate 5×10⁻⁵, menghasilkan nilai Accuracy 0,8538, Precision 0,8539, Recall 0,8538,dan F1-Score 0,8504. Penelitian ini membuktikan bahwa NusaBERT yang telah dilatih padakorpus bahasa daerah Indonesia mampu menangani kompleksitas linguistik teks code-mixedIndonesia-Jawa secara efektif.
Analisis Sentimen Lexicon-Based Penggunaan ChatGPT pada Siswa dan Guru SMAN 2 Klari Bunga Khoeriah Utami; Sofi Defiyanti; Mohamad Jajuli
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7153

Abstract

The development of Artificial Intelligence (AI) technology, particularly ChatGPT, has increasingly been utilized in learning activities. However, the perceptions of students and teachers regarding the use of ChatGPT in schools still need to be objectively mapped. This study aims to analyze the sentiments of students and teachers at SMAN 2 Klari toward the use of ChatGPT in learning activities using a Lexicon-based approach with the InSet sentiment lexicon. The research employed a quantitative approach using 777 responses collected from 742 students and 35 teachers through open-ended questionnaires. The research stages included data collection, preprocessing (case folding, text cleaning, tokenizing, stopword removal, slang word normalization, and stemming), Lexicon-based sentiment analysis, and evaluation using Fleiss’ Kappa and Confusion Matrix. The results showed that the majority of students expressed positive sentiment toward the use of ChatGPT in learning, accounting for 91.1%, while negative and neutral sentiments accounted for 6.9% and 2.0%, respectively. Among teachers, all respondents expressed positive sentiment, reaching 100%. The annotator evaluation using Fleiss’ Kappa obtained a score of 0.7745, categorized as Substantial Agreement, indicating strong agreement among annotators. Furthermore, the Confusion Matrix evaluation produced an accuracy of 79%, demonstrating that the Lexicon-based method performed reasonably well in classifying sentiments related to ChatGPT usage. Overall, the findings indicate that ChatGPT is perceived positively by both students and teachers as a technology that supports the learning process, although concerns remain regarding potential dependency and inappropriate use.
Analisis Sentimen Ulasan Skintific Aqua Light Daily Sunscreen di Tokopedia Menggunakan InSet Lexicon Nadiyah Nur Rafifah; Mohammad Jajuli; Sofi Defiyanti
IKRA-ITH Informatika : Jurnal Komputer dan Informatika Vol. 10 No. 2 (2026): IKRAITH-INFORMATIKA Vol 10 No 2 Juli 2026
Publisher : Fakultas Teknik Universitas Persada Indonesia YAI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.37817/ikraith-informatika.v10i2.7210

Abstract

Customer reviews on e-commerce platforms can be utilized to understand consumer perceptions of a product through sentiment analysis. This study aims to analyze the sentiment of customer reviews for Skintific Aqua Light Daily Sunscreen on the Tokopedia platform using the lexicon-based method with the InSet Lexicon and to evaluate its classification performance. The study applies the Knowledge Discovery in Databases (KDD) framework, which consists of data selection, data preprocessing, data transformation, data mining, and evaluation. The data were collected through web scraping using Python. Of the 672 reviews collected, 634 reviews remained after the data cleaning process. The preprocessing stage included case folding, text normalization, tokenization, stopword removal, and stemming. The results showed that 532 reviews (83.91%) were classified as positive, 78 reviews (12.30%) as negative, and 24 reviews (3.79%) as neutral. The evaluation was conducted on 100 randomly selected review samples validated by two validators. The inter-rater agreement, measured using Cohen's Kappa, was 0.34, indicating a Fair Agreement. Furthermore, the evaluation using a confusion matrix yielded an accuracy of 64.00%, a precision of 54.71%, a recall of 64.00%, and an F1-score of 58.42%. These results indicate that the lexicon-based method using the InSet Lexicon is capable of classifying most customer review sentiments, although it still has limitations in identifying neutral sentiment.
PENERAPAN MODEL XGBOOST DENGAN INTERPRETASI MENGGUNAKAN SHAP UNTUK DETEKSI AKTIVITAS PENCUCIAN UANG Joyce Rosita Firdaus; Sofi Defiyanti; Betha Nurina Sari
Jurnal Informatika dan Teknik Elektro Terapan Vol. 14 No. 3 (2026)
Publisher : Universitas Lampung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.23960/jitet.v14i3.9993

Abstract

Deteksi pencucian uang pada data transaksi keuangan sangat kompleks karena ketidakseimbangan kelas yang ekstrem, sehingga diperlukan model yang akurat sekaligus dapat dijelaskan. Penelitian ini menerapkan algoritma eXtreme Gradient Boosting (XGBoost) untuk mengklasifikasikan transaksi normal dan transaksi laundering, serta menginterpretasikan hasil prediksi dengan SHapley Additive exPlanations (SHAP). Metode penelitian yang digunakan adalah CRISP-DM, meliputi business understanding, data understanding, data preparation, modeling, evaluation, dan deployment. Data yang digunakan merupakan dataset transaksi keuangan sintetis sebanyak 9.504.852 transaksi dengan proporsi transaksi laundering sebesar 0,10%. Feature engineering menghasilkan 25 fitur yang merepresentasikan pola transaksi, hubungan antar akun, frekuensi, dan rasio transaksi. Tiga pendekatan penanganan data tidak seimbang diuji: SMOTE, scale_pos_weight, dan kombinasi keduanya. Hasil terbaik diperoleh dengan XGBoost menggunakan scale_pos_weight, dengan accuracy 0,9996, precision 0,7796, recall 0,9018, dan F1-score 0,8363. Analisis SHAP menunjukkan receiver_fan_in dan sender_fan_out sebagai fitur paling berpengaruh dengan kontribusi kumulatif sebesar 63,09%, menegaskan bahwa pola hubungan antar akun menjadi indikator utama dalam deteksi pencucian uang.
Co-Authors Ade Andri Hendriadi Agung Susilo Yuda Ahmad Khusaeri Alfajr, Nur Halizah Alia Fadilah Angelina Raharjo, Fadya Aries Suharso Betha Nurina Sari Billy Ibrahim Hasbi Budi Arif Dermawan Bunga Khoeriah Utami Carudin Dadang Yusup garno, Garno Hannie, Hannie Intan Purnamasari Iqbal Maulana Jaman, Jajam Haerul Joyce Rosita Firdaus M Jajuli Mayasari, Rini Mohammad Jajuli Mohammad Nurfaizy Pangestu Muhamad Alghivary, Mirza Muhamad Rizky Abadi Muhammad Fadhlil Aziz Nadiyah Nur Rafifah Nina Sulistiyowati Nur Padilah, Tesa Nur Rachman, Muhammad Hafidz Nurul Rohmawati Nurul Rohmawati W Nurul Rohmawati, Nurul Padilah, Tesa Nur Purnamasari , Intan Ratna Mufidah Rini Mayasari Safitri, Nur Sopian Syauri Supriyanto Lumbanbatu Syafi'i, Muhammad Taufik Ridwan Tri Isna Kuswaya Wikarta, Nana Casmana Ade Yanti, Fajri Yuyun Umaidah