Yunansah, Fakhri Sabran (2026) Integrasi BERT dan Named Entity Recognition untuk Analisis Emosi Multilabel Komentar Pers Presiden di Youtube. Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (Cover)
22082010153.-cover.pdf Download (1MB) | Preview |
|
|
Text (Bab 1)
22082010153.-bab1.pdf Download (761kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 2)
22082010153.-bab2.pdf Restricted to Repository staff only until 25 May 2028. Download (1MB) |
|
|
Text (Bab 3)
22082010153.-bab3.pdf Restricted to Repository staff only until 25 May 2028. Download (948kB) |
|
|
Text (Bab 4)
22082010153.-bab4.pdf Restricted to Repository staff only until 25 May 2028. Download (7MB) |
|
|
Text (Bab 5)
22082010153.-bab5.pdf Download (129kB) | Preview |
|
|
Text (Daftar pustaka)
22082010153.-daftarpustaka.pdf Download (128kB) | Preview |
|
|
Text (Lampiran)
22082010153.-lampiran.pdf Restricted to Repository staff only Download (1MB) |
Abstract
The rapid growth of social media, particularly YouTube, has made it a majorplatform for the public to express opinions and emotions regarding various publicissues, including of icial government statements. Comments on IndonesianPresidential press conference videos reflect complex emotional expressions that often appear simultaneously within a single text. This condition requires ananalytical approach capable of handling multilabel characteristics whileunderstanding dynamic language contexts. This study aims to analyze theperformance of integrating Bidirectional Encoder Representations fromTransformers (BERT) with Named Entity Recognition (NER) in multilabel emotion classification on YouTube comments using five emotion categories: love, happiness, anger, fear, and sadness. The method used in this study is a pipeline- based integration approach, where NER is utilized to enrich text representationbefore classification using IndoBERT. The dataset was obtained frompubliccomments on a Presidential press conference video uploaded to the of icial YouTube channel of the Indonesian Presidential Secretariat. The results showthat NER integration improves model performance in several scenarios. The best model was achieved using back-translation augmentation with NER integrationunder the 80:20 data split scenario, resulting in an F1-score (micro) of 0.7843and a Hamming Loss of 0.1199. The model achieved the best performance byproviding a balanced precision and recall through data augmentation and entityinformation. NER integration and data augmentation improved emotionclassification performance for Indonesian public comments.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | H Social Sciences > HG Finance > HG1709 Data processing | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Information Systems | ||||||||||||
| Depositing User: | Fakhri Yunansah | ||||||||||||
| Date Deposited: | 26 May 2026 02:13 | ||||||||||||
| Last Modified: | 26 May 2026 03:52 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/52523 |
Actions (login required)
 |
View Item |