Text

Analisis sentimen opini publik terhadap layanan BSSN dengan menggunakan BERT model

---

Badan Siber dan Sandi Negara atau BSSN merupakan merupakan instansi pelayanan siber dan sandi di Indonesia yang memiliki tanggung jawab meliputi keamanan di bidang siber dan persandian. Adapun layanan yang diberikan BSSN beberapa diantaranya layanan kerjasama sertifikat elektronik, Aplikasi Password Manager Satria dan Aplikasi BeSign. Untuk mengetahui opini publik sekaligus sebagai timbal balik dari masyarakat terkait layanan BSSN, penelitian ini telah dilakukan analisis sentimen pada media sosial twitter. Teks yang diunggah pada media sosial twitter tidak memiliki aturan khusus dalam penulisannya sehingga menimbulkan ragam variasi bahasa tidak baku (slangword) sehingga pada penelitian ini akan menerapkan penanganan slangword pada teks sebelum diproses dalam pemodelan. Digunakan dua buah model analisis sentimen yang dapat menangani analisis teks berbahasa Indonesia, yakni BERT-base Multilingual Model dan IndoBERT-base Model. Kedua model tersebut dibandingkan dan hasilnya IndoBERT-base Model lebih baik dibandingkan dengan BERT-base Multilingual Model. Model terbaik IndoBERT-base Model memiliki akurasi sebesar 0,70 yang kemudian model ini diimplementasikan dalam bentuk dashboard menggunakan framework Flask Python. --

The National Cyber and Crypto Agency or BSSN is a cyber and crypto service agency in Indonesia that has responsibilities covering cyber and encryption security. Some of the services provided by BSSN include electronic certificate collaboration services, the Satria Password Manager Application and the BeSign Application. To find out public opinion as well as feedback from the community regarding BSSN services, this research has conducted a sentiment analysis on Twitter social media. The text uploaded on Twitter social media does not have special rules in its writing, giving rise to a variety of non-standard language variations (slangwords) so that in this study we will apply slangword handling to the text before it is processed in modeling. Two sentiment analysis models are used that can handle the analysis of Indonesian texts, namely the BERT-base Multilingual Model and the IndoBERT Model. The two models were compared and the result was that the IndoBERT-base Model was better than the BERT-base Multilingual Model. The best model IndoBERT-base Model has an accuracy of 0.70 which is then implemented in the form of a dashboard using the Flask Python framework.

---

Availability

Detail Information

Series Title

-

Call Number

2022 ATI a

Publisher

Bogor : Politeknik Siber dan Sandi Negara., 2022

Collation

xvi, 69 hlm.

Language

Indonesia

ISBN/ISSN

--

Classification

--

Content Type

-

Media Type

-

Carrier Type

-

Edition

--

Subject(s)

Analisis Sentimen
BERT Multilingual Model
BSSN
IndoBERT Model

Specific Detail Info

-

Statement of Responsibility

Atika Nurliana

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment