SISKA, LESTARI (2025) IMPLEMENTASI DEEP LEARNING UNTUK ANALISIS SENTIMEN PADA DATA X DALAM PREDIKSI TREN “KOREAN STYLE”. Diploma thesis, UIN Raden Intan Lampung.
|
PDF
Restricted to Repository staff only Download (4MB) |
|
|
PDF
Restricted to Repository staff only Download (5MB) |
Abstract
ABSTRAK Penelitian ini bertujuan untuk mengembangkan dan mengevaluasi model deep learning berbasis Convolutional Neural Network (CNN) guna melakukan analisis sentimen terhadap tweet berbahasa Indonesia yang berkaitan dengan tren “Korean Style”. Fenomena ini menunjukkan peningkatan diskusi publik di media sosial, khususnya X, seiring populernya budaya Korea di Indonesia. Metode yang digunakan mencakup pengumpulan data sebanyak 7.187 tweet melalui proses web crawling dengan kata kunci seperti “Korea Style”, “K-pop”, dan “Korean fashion”. Data tersebut melalui tahapan pra-pemrosesan teks, meliputi case folding, penghapusan karakter khusus dan emoji, stopword removal, tokenisasi, dan lemmatization. Untuk merepresentasikan teks ke dalam vektor numerik, digunakan dua teknik word embedding: Word2Vec dan FastText. Model CNN kemudian dilatih untuk mengklasifikasikan sentimen ke dalam tiga kategori: positif, negatif, dan netral. Evaluasi kinerja dilakukan menggunakan metrik akurasi, precision, recall, F1-score, serta confusion matrix dan grafik visualisasi proses pembelajaran. Hasil penelitian menunjukkan bahwa model CNN mampu mencapai akurasi validasi sebesar 71,26%, dengan nilai F1-score tertinggi sebesar 0,81 pada kelas netral. Namun, performa pada kelas negatif masih rendah akibat ketidakseimbangan data antar kelas. Analisis menunjukkan bahwa Word2Vec memberikan hasil lebih stabil dibanding FastText dalam konteks ini. Penelitian ini berkontribusi pada literatur dengan mengaplikasikan CNN secara spesifik pada tweet berbahasa Indonesia terkait tren budaya populer, serta menunjukkan tantangan dan peluang dalam menangani ketidakseimbangan kelas. Rekomendasi pengembangan model selanjutnya mencakup penggunaan teknik data augmentation, metode loss yang adaptif (seperti focal loss), serta eksplorasi model transformer seperti IndoBERT untuk peningkatan akurasi klasifikasi. Kata Kunci: Analisis Sentimen, Convolutional Neural Network (CNN), Korea Style ii. ABSTRACT This study aims to develop and evaluate a deep learning model based on Convolutional Neural Network (CNN) for sentiment analysis of Indonesian-language tweets related to the “Korea Style” trend. This phenomenon has gained significant attention on social media, especially X, alongside the rising popularity of Korean culture in Indonesia. A total of 7,187 tweets were collected through web crawling using keywords such as “Korean Style”, “K-pop”, and “Korean fashion”. The dataset underwent a comprehensive text preprocessing pipeline, including case folding, removal of special characters and emojis, stopword elimination, tokenization, and lemmatization. Two word embedding techniques—Word2Vec and FastText—were employed to convert text into numerical vectors. The CNN model was then trained to classify sentiment into three categories: positive, negative, and neutral. Model performance was evaluated using accuracy, precision, recall, F1-score, confusion matrix, and visualizations of training accuracy and loss. The results indicated that the CNN model achieved a validation accuracy of 71.26%, with the highest F1-score of 0.81 for the neutral class. However, classification performance for negative sentiment remained low due to class imbalance in the dataset. Analysis showed that Word2Vec produced more stable results compared to FastText in this specific task. This research contributes to the literature by implementing CNN for sentiment classification of Indonesian tweets on a specific cultural trend and highlights challenges in handling class imbalance. Future improvements may involve text data augmentation, adaptive loss functions (e.g., focal loss), and exploration of transformer-based models such as IndoBERT to enhance classification performance. Keywords: Sentiment Analysis, Convolutional Neural Network (CNN), Korean Style iii.
| Item Type: | Thesis (Diploma) |
|---|---|
| Subjects: | Sistem Informasi |
| Divisions: | Fakultas Sains dan Teknologi > Sistem Informasi |
| Depositing User: | LAYANAN PERPUSTAKAAN UINRIL REFERENSI |
| Date Deposited: | 09 Oct 2025 04:06 |
| Last Modified: | 09 Oct 2025 04:06 |
| URI: | https://repository.radenintan.ac.id/id/eprint/40971 |
Actions (login required)
![]() |
View Item |


