Skripsi

KLASIFIKASI EMOSI MULTI-LABEL PADA TEKS BERBAHASA INDONESIA DENGAN FINE-TUNING INDOBERT MENGGUNAKAN DATASET GoEmotions

---

Penilaian

---

The rapid growth of social media interactions in Indonesia has generated a massive volume of text data containing diverse emotional expressions. The linguistic complexity and contextual richness of the Indonesian language make emotion identification and classification a challenging task. To address this issue, this study develops a multi-label emotion classification system by applying fine-tuning on the IndoBERT model, a pre-trained transformer specifically designed to understand Indonesian syntax and semantics. The dataset used in this study is a translated version of GoEmotions, consisting of 58,000 Reddit comments annotated with 28 emotion categories. The fine-tuning process was conducted through six experimental scenarios, combining different learning rates (2e-6, 4e-6, 5e-5) and batch sizes (16 and 32). Model performance was evaluated using Exact Match, Precision Macro, Recall Macro, and F1-Score Macro. The best results were achieved with a learning rate of 4e-6 and batch size of 32, yielding an Exact Match of 42.27% and an F1-Score Macro of 37.59%, while the configuration with a learning rate of 5e-5 and batch size of 16 achieved the highest F1-Score of 39.11% but exhibited signs of overfitting. These findings demonstrate that the fine-tuned IndoBERT model effectively performs multi-label emotion classification on Indonesian texts and can serve as a strong baseline for future emotion analysis and Natural Language Processing (NLP) research in the Indonesian language. Keywords: Multi-Label Emotion Classification, IndoBERT, GoEmotions, Fine-Tuning, Exact Match

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2507006228	T185574	T1855742025	Central Library (Reference)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1855742025

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xiv, 111 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

004.07

Content Type

Text

Media Type

unmediated

Carrier Type

other (computer)

Edition

-

Subject(s)

Prodi Teknik Informatika
Teknik Pengelolaan Berita

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment