IDENTIFIKASI KOMENTAR SPAM DI FACEBOOK MENGGUNAKAN LONG SHORT-TERM MEMORY (LSTM)

SILANGIT, AMOS AUGUSTO and Abdiansah, Abdiansah and Al Farissi, Al Farissi (2023) IDENTIFIKASI KOMENTAR SPAM DI FACEBOOK MENGGUNAKAN LONG SHORT-TERM MEMORY (LSTM). Undergraduate thesis, Sriwijaya University.

	Text RAMA_55201_09021281924051.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (8MB) | Request a copy
	Text RAMA_55201_09021281924051_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (4MB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (3MB)
	Text RAMA_55201_09021281924051_0001108401_0215128501_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (507kB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (68kB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (4MB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (697kB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_06.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (10kB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_07_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (142kB) | Request a copy
	Text RAMA_55201_09021281924051_0001108401_0215128501_08_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (4kB) | Request a copy

Abstract

Komentar spam di platform sosial media, khususnya Facebook, menjadi permasalahan yang semakin meruncing seiring dengan pertumbuhan pengguna dan interaksi online. Penelitian ini bertujuan untuk mengembangkan identifikasi komentar spam menggunakan pendekatan Long Short-Term Memory (LSTM), sebuah model jaringan saraf rekuren yang mampu memahami konteks temporal dalam data teks. Penelitian ini melibatkan pengumpulan data komentar dari Facebook, dan LSTM digunakan untuk mengekstrak pola temporal yang terkandung dalam setiap komentar. Evaluasi dilakukan terhadap keefektifan metode ini dalam membedakan antara komentar spam dan non-spam, dengan memperhatikan aspek-aspek kinerja seperti akurasi, presisi, recall, dan F-measure. Data yang digunakan didapatkan dari komentar-komentar suatu post di Facebook sebanyak 800 data komentar. Pengujian menggunakan 6 skenario percobaan hasil pengujian terbaik diperoleh dengan hyper parameter nilai layer dropout sebesar 0, jumlah neuron LSTM sebesar 64, nilai learning rate sebesar 10-3, nilai dropoutLSTM sebesar 0.5, nilai batch size sebesar 8, dan nilai epoch sebesar 10. Hasil didapatkan dengan nilai accuracy sebesar 87.50%, precision sebesar 86.98%, recall sebesar 87.55%, dan f-measure sebesar 87.21%.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Spam, Identifikasi, Facebook, Long Short-Term Memory
Subjects:	T Technology > T Technology (General) > T57.6-57.97 Operations research. Systems analysis
Divisions:	09-Faculty of Computer Science > 55201-Informatics (S1)
Depositing User:	Amos Augusto Silangit
Date Deposited:	26 Feb 2024 01:42
Last Modified:	26 Feb 2024 01:42
URI:	http://repository.unsri.ac.id/id/eprint/141050

Actions (login required)

View Item