ANALISIS SENTIMEN MASYARAKAT TERHADAP FILM “DIRTY VOTED” DI YOUTUBE MENGGUNAKAN MODEL SVM (SUPPORT VECTOR MACHINE) DAN LSTM (LONG SHORT-TERM MEMORY)

ARDIANSYAH, MUHAMAD RIZKY (2024) ANALISIS SENTIMEN MASYARAKAT TERHADAP FILM “DIRTY VOTED” DI YOUTUBE MENGGUNAKAN MODEL SVM (SUPPORT VECTOR MACHINE) DAN LSTM (LONG SHORT-TERM MEMORY). S1 thesis, Universitas Mercu Buana - Menteng.

[img] Text (Cover)
41520110050-MUHAMAD RIZKY ARDIANSYAH-01 COVER - Muhamad Rizky Ardiansyah.pdf
Download (501kB)
[img] Text (Abstrak)
41520110050-MUHAMAD RIZKY ARDIANSYAH-02 ABSTRAK - Muhamad Rizky Ardiansyah.pdf
Download (177kB)
[img] Text (Bab 1)
41520110050-MUHAMAD RIZKY ARDIANSYAH-03 Bab 1 - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (252kB)
[img] Text (Bab 2)
41520110050-MUHAMAD RIZKY ARDIANSYAH-04 Bab 2 - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (315kB)
[img] Text (Bab 3)
41520110050-MUHAMAD RIZKY ARDIANSYAH-05 Bab 3 - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (356kB)
[img] Text (Bab 4)
41520110050-MUHAMAD RIZKY ARDIANSYAH-06 Bab 4 - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (5MB)
[img] Text (Bab 5)
41520110050-MUHAMAD RIZKY ARDIANSYAH-07 Bab 5 - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (228kB)
[img] Text (Daftar pustaka)
41520110050-MUHAMAD RIZKY ARDIANSYAH-09 Daftar Pustaka - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (220kB)
[img] Text (Lampiran)
41520110050-MUHAMAD RIZKY ARDIANSYAH-10 Lampiran - Muhamad Rizky Ardiansyah.pdf
Restricted to Registered users only
Download (589kB)
[img] Text (Formulir kebasahan dan publikasi TA)
Persetujuan Publikasi Perpus - Muhamad Rizky Ardiansyah.pdf
Restricted to Repository staff only
Download (97kB)

Abstract

Film dokumenter "Dirty Vote" yang tersedia di kanal YouTube PSHK Indonesia, telah menjadi salah satu video yang memperoleh perhatian dan reaksi yang beragam di platform YouTube. Analisis sentimen dilakukan untuk mengetahui sentimen masyarakat dalam memberikan komentar terhadap film dokumenter ini. Dalam menganalisis sentimen masyarakat, digunakan InSet Lexicon sebagai kamus untuk proses pelabelan otomatis bahasa Indonesia, SVM dan LSTM sebagai algoritma klasifikasi sentimen dan SMOTE untuk mengatasi overfitting. Penelitian ini melakukan perbandingan hasil pelabelan dan model di tiap tahap pra-pemrosesan teks. Hasil penelitian menunjukkan penggunaan InSet Lexicon untuk pelabelan pada tahap terakhir di pra-pemrosesan stemming menghasilkan distribusi sentimen yang paling timpang, sentimen positif sebesar 20,26% dan negatif mencapai 79,74%. Selain itu, hasil evaluasi dari kinerja SVM mencatat akurasi tertinggi sebesar 97,58% menggunakan data di proses stemming dengan test size 10% dan random state 1, sementara LSTM mencatat akurasi tertinggi sebesar 96,3% dan nilai loss function yaitu 0,098 dengan batch size 16 dengan menggunakan data ditahap stemming. Penggunaan metode SMOTE untuk mengatasi overfitting menunjukkan hasil yang baik untuk model SVM sedangkan hasil yang negatif untuk LSTM. The documentary "Dirty Vote" on PSHK Indonesia's YouTube channel has become one of the videos that has gained attention and mixed reactions on the YouTube platform. Sentiment analysis was conducted to find out people's sentiments in commenting on this documentary. In analyzing public sentiment, InSet Lexicon is used as a dictionary for the automatic labeling process of Indonesian language, SVM and LSTM as sentiment classification algorithms and SMOTE to overcome overfitting. This research compares the labeling results and models at each stage of text pre-processing. The results showed that the use of InSet Lexicon for labeling at the last stage of stemming pre-processing resulted in the most unequal sentiment distribution, with positive sentiment accounting for 20,26% and negative sentiment accounting for 79,74%. In addition, the evaluation results of SVM performance recorded the highest accuracy of 97,58% using data in the stemming process with test size 10% and random state 1, while LSTM recorded the highest accuracy of 96,3% and loss function value of 0,098 with batch size 16 using data in the stemming stage. The use of the SMOTE method to overcome overfitting showed good results for the SVM model while negative results for the LSTM.

Item Type: Thesis (S1)
NIM/NIDN Creators: 41520110050
Uncontrolled Keywords: Analisis sentimen, SVM, LSTM, SMOTE, Lexicon Sentiment Analysis, SVM, LSTM, SMOTE, Lexicon
Subjects: 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 004 Data Processing, Computer Science/Pemrosesan Data, Ilmu Komputer, Teknik Informatika
Divisions: Fakultas Ilmu Komputer > Informatika
Depositing User: WIDYA AYU PUSPITA NINGRUM
Date Deposited: 19 Aug 2024 04:44
Last Modified: 19 Aug 2024 04:44
URI: http://repository.mercubuana.ac.id/id/eprint/90322

Actions (login required)

View Item View Item
UMB Repository is powered by EPrints 3 which is developed by the School of Electronics and Computer Science at the University of Southampton. More information and software credits.