Komparasi Metode Label Powerset K-NN dan ML-KNN dalam Klasifikasi Multi-label Cyberbullyng pada Komentar Instagram

Fadlilah, Imamah Nur (2025) Komparasi Metode Label Powerset K-NN dan ML-KNN dalam Klasifikasi Multi-label Cyberbullyng pada Komentar Instagram. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21082010067-cover.pdf Download (961kB)
	Text (Bab 1) 21082010067-bab1.pdf Download (55kB)
	Text (Bab 2) 21082010067-bab2.pdf Restricted to Repository staff only until 19 March 2027. Download (184kB)
	Text (Bab 3) 21082010067-bab3.pdf Restricted to Repository staff only until 19 March 2027. Download (265kB)
	Text (Bab 4) 21082010067-bab4.pdf Restricted to Repository staff only until 19 March 2027. Download (5MB)
	Text (Bab 5) 21082010067-bab5.pdf Download (11kB)
	Text (Daftar pustaka) 21082010067-daftarpustaka.pdf Download (91kB)
	Text (Lampiran) 21082010067-lampiran.pdf Restricted to Repository staff only until 19 March 2027. Download (1MB)

Abstract

Cyberbullying is a form of bullying conducted through digital media, including social media platforms such as Instagram. With its high number of users and interactive comment features, Instagram is one of the platforms with the highest incidence of cyberbullying. The data in this study was obtained from comments on public figure accounts. This research aims to apply multi-label classification in detecting cyberbullying on Instagram comments. Two main approaches were compared, namely Problem Transformation with Label Powerset KNN and Algorithm Adaptation with ML-KNN. The research data was converted using TF-IDF feature extraction technique with a combination of n-grams (1-3) to improve the model accuracy. The results showed that ML-KNN performed better than Label Powerset KNN. ML-KNN shows a higher F1-score of 0.849 compared to LP-KNN of 0.828, and a lower hamming loss of 0.110 compared to 0.124 in LP-KNN. Thus, ML-KNN is more accurate in handling multi-label data. In addition, the developed system can classify comments into various cyberbullying categories simultaneously, supports input in the form of single text as well as CSV files, thus enabling large-scale analysis. With these findings, the research is expected to contribute to the development of cyberbullying detection systems to enhance digital safety on social media.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Wahyuni, Eka Dyar	NIDN0001128406	ekawahyuni.si@upnjatim.ac.id
Thesis advisor	Permatasari, Reisa	NIDN0014059203	reisa.permatasari.sifo@upnjatim.ac.id

Subjects:

Q Science > Q Science (General)
Q Science > QA Mathematics
Q Science > QA Mathematics > QA76.6 Computer Programming
T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Information Systems

Depositing User:

IMAMAH FADLILAH

Date Deposited:

19 Mar 2025 03:54

Last Modified:

19 Mar 2025 03:54

URI:

https://repository.upnjatim.ac.id/id/eprint/35873