Skripsi

FINE-TUNING INDOBERT UNTUK KLASIFIKASI KATEGORI BERITA BERBAHASA INDONESIA.

---

The availability of Indonesian news articles on the internet has greatly increased, making it more challenging to recognize and categorize news accurately. Therefore, a solution to this issue is to develop a classification system for Indonesian news article categories. This research aims to classify Indonesian news category using fine-tuning on the pre-trained IndoBERT model. The dataset consists of 31,993 articles divided into five news categories: education, health, technology, sports, and automotive. Articles were collected from two of the largest and most trusted online news portals, kompas.com and detik.com, using web scraping method. The fine-tuning process was divided into 8 scenarios, which are combinations of dataset type configurations, learning rate, and batch size. Based on the test results, the highest accuracy was obtained in scenario 2, where the model trained with a learning rate of 2e-5 and batch size of 32, reaching an accuracy of 98.37%.

---

Availability

Detail Information

Series Title

-

Call Number

T1627362024

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2024

Collation

xvii, 97 hlm.; ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

004.07

Content Type

Text

Media Type

unmediated

Carrier Type

other (computer)

Edition

-

Subject(s)

Pemrosesan data
Prodi Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

MURZ

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment