Skripsi

KLASIFIKASI EMOSI PADA LIRIK LAGU BERBAHASA INDONESIA DENGAN FINE-TUNING INDOBERT

---

The rapid growth of digital music platforms and the increasing availability of Indonesian song lyrics have created a demand for automatic emotion identification to support content analysis and mood-based recommendation. However, emotion classification in song lyrics is challenging due to subjective interpretation, figurative language, and class imbalance across emotion categories. This study aims to develop an emotion classification system for Indonesian song lyrics by fine-tuning the IndoBERT model. The dataset was compiled from Kaggle and Genius.com web scraping, and labeled using a human-in-the-loop scheme combining an AI ensemble and manual verification. To mitigate class imbalance, 250 synthetic (augmented) lyrics were generated for minority emotion classes. The final dataset consists of 1,397 lyrics across four emotion classes—happy, sad, angry, and fear—and is split into 80% training, 10% validation, and 10% testing sets. Experiments were conducted using 12 hyperparameter configurations (learning rate, batch size, and weight decay). Model performance was primarily evaluated using Macro F1-score, with accuracy, precision, and recall as supporting metrics. The best performance was achieved by Configuration 7 (identical to Configuration 8) with a learning rate of 2e-05, batch size of 16, and weight decay of 0.01, reaching a Macro F1-score of 0.6505 and an accuracy of 0.6522. Additionally, varying weight decay within 0.01–0.05 did not yield a significant performance difference in this setting.

---

Availability

Detail Information

Series Title

-

Call Number

T1899322025

Publisher

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

Collation

xiv, 126 hlm.; ilus.; tab.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.320 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Pemrosesan Bahasa Alami
Prodi Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment