Skripsi

PENERAPAN FINE-TUNING INDOBERT UNTUK STANCE DETECTION PADA KOMENTAR PEMBATALAN KONSER HINDIA DI TIKTOK

---

TikTok has evolved into an arena for polarized public opinion, particularly concerning the cancellation of musician Hindia's concert due to issues of satanic symbolism. This study applies the IndoBERT P2 model for stance detection on 2,306 Indonesian comments collected from June 18 to July 20, 2025. Challenges involving noisy data and class imbalance were addressed using the Random Oversampling technique and an Early Stopping mechanism to prevent overfitting. The analysis results reveal a dominance of negative responses, with the Against class reaching 69.6%, followed by Favor (24.9%) and None (5.5%). Model evaluation yielded a Macro F1-Score of 0.60, with the highest performance observed in the Against class (F1 0.83). However, performance in the Favor (0.62) and None (0.33) classes remains limited due to the model's difficulty in comprehending sarcasm. These findings present an initial benchmark for IndoBERT P2 performance and offer strategic insights for artist management in crisis mitigation.

---

Availability

Detail Information

Series Title

-

Call Number

T1894012025

Publisher

Indralaya : Prodi Sistem Informasi, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xvi, 88 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

006.350 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Informasi
Penerapan Fine-Tuning

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment