Skripsi

ANALISIS SENTIMEN BERBASIS TRANSFORMER (INDOBERT) PADA ULASAN PENGGUNA APLIKASI GOJEK DI GOOGLE PLAY STORE DENGAN SKEMA 3-KELAS DAN 5-KELAS

---

This study implements and compares the performance of the IndoBERT deep learning model in analyzing the sentiment of Gojek application reviews using 3-class and 5-class schemes. Data was extracted via web scraping, yielding 66,709 raw reviews, which were then preprocessed and balanced into 23,115 (3-class) and 32,899 (5-class) datasets. Training was executed with an 80:10:10 ratio across six scenarios varying the learning rate and epochs. Experimental results indicate the 5-class scheme achieved an optimal accuracy of 98.69% (learning rate 2e-5, 8 epochs), consistently outperforming the 3-class scheme (97.62%) under similar configurations. Further testing proved that the elimination of preprocessing degraded model accuracy by 20-27%. In conclusion, the 5-class labeling scheme facilitates IndoBERT in mapping sentiment polarity more comprehensively, while preprocessing is proven to be an absolute prerequisite in handling informal text data.

---

Availability

Detail Information

Series Title

-

Call Number

T1955952026

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xix, V-9 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

006.350 7

Content Type

Text

Media Type

unmediated

Carrier Type

-

Edition

-

Subject(s)

Indobert
 Prodi Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

KA

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment