Rianto, Rizky (2026) Penerapan Metode Rocchio Untuk Klasifikasi Buzzer Di Media Sosial Instagram (Study Kasus Demo Kenaikan Tunjangan Anggota Dpr). Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
20081010237.-Cover.pdf Download (1MB) |
|
Text (Bab 1)
20081010237.-Bab 1.pdf Download (91kB) |
|
Text (Bab 2)
20081010237.-Bab 2.pdf Restricted to Repository staff only until 28 January 2029. Download (178kB) |
|
Text (Bab 3)
20081010237.-Bab 3.pdf Restricted to Repository staff only until 28 January 2029. Download (4MB) |
|
Text (Bab 4)
20081010237.-bab 4.pdf Restricted to Repository staff only until 28 January 2029. Download (1MB) |
|
Text (Bab 5)
20081010237.-bab 5.pdf Download (78kB) |
|
Text (Daftar Pustaka)
20081010237.-daftarpustaka.pdf Download (83kB) |
|
Text (Lampiran)
20081010237.-lampiran.pdf Restricted to Repository staff only until 28 January 2029. Download (158kB) |
Abstract
Coordinated buzzer activities on Instagram social media have the potential to manipulate public opinion, especially on sensitive issues such as demonstrations against the increase in allowances for members of the House of Representatives. Manual detection of buzzer comments becomes inefficient given the large volume of data. Therefore, this study aims to implement the Rocchio (Nearest Centroid) method to build an automatic classification system that is able to categorize Instagram comments into three classes: Pro buzzer (support), Contra buzzer (against), and Non-buzzer (neutral). The research data was obtained through Instagram comment crawling using the keywords "#tunjangandpr" and "#demodpr" in the period of August 19–30, 2025, resulting in 3,745 raw comments. A total of 3,457 comments were then labeled through an automated sentiment analysis approach with the VARDER library. The preProcessing stage includes cleaning, case folding, tokenizing, stopword removal, and lemmatization. Representation of text features using TF-IDF (Term Frequency-Inverse Document Frequency) weighting. The Rocchio model was evaluated with five holdout data sharing scenarios (90:10, 80:20, 70:30, 60:40, 50:50) and measured using a confusion matrix and accuracy, precision, recall, and F1-score metrics. The results of the experiments showed that the 90:10 scenario produced the best performance with an accuracy of 82%, macro precision of 0.82, macro recall of 0.82, and F1-score of macro 0.82. The model does best at recognizing the Pro buzzer class (0.96 recall), followed by the Contra buzzer, while the Non-buzzer class has a lower recall (0.72) due to the wider variety of neutral language. In addition to comment-level classification, this study developed an account analysis feature that groups comments by username to identify the main tendencies of an account. To handle the tie condition, a word-count tie-breaker mechanism is applied. The conclusions of the study confirm that the Rocchio method, although simple, is effective and efficient as a baseline model for the classification of three classes on Instagram comment data with sparse characteristics such as TF-IDF. The developed account analysis provides added value by offering an aggregate perspective of user behavior. These results contribute to the field of computational social science by providing a fast and implementable buzzer detection method, as well as being the foundation for the development of a more sophisticated public opinion monitoring system on social media platforms. Keywords: Rocchio, text classification, buzzer, Instagram, TF-IDF, social media comments.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics > QA76 Computer software Q Science > QA Mathematics > QA76.6 Computer Programming |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Rizky Rizky Rianto Rizky | ||||||||||||
| Date Deposited: | 29 Jan 2026 07:54 | ||||||||||||
| Last Modified: | 29 Jan 2026 08:13 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/49232 |
Actions (login required)
 |
View Item |