PERBANDINGAN METODE LONG SHORT-TERM MEMORY DENGAN CONVOLUTIONAL NEURAL NETWORK DALAM ANALISIS SENTIMEN

• Main Author: Aldilla Gardika Pramesta, (Author)
• Format: Book
• Published: 2022-07-21.
• Subjects: Thesis; NonPeerReviewed
• Online Access: Link Metadata

Summary

One of the algorithms that can be used to analyze sentiment is Deep Learning. Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) are deep learning algorithms that are quite popular and suitable for use in sentiment analysis. In this study, a comparative analysis of the LSTM method with CNN was conducted in analyzing sentiment so as to obtain more optimal performance results in analyzing sentiment. The data for this research are tweets obtained from the Kaggle website, which is a dataset on sentiment towards television broadcasts in Indonesia. The final result of this research is that the two methods tend to be balanced. The highest accuracy obtained by the LSTM method is 85% with a loss value of 0.0022 and the lowest accuracy is 78% and an average accuracy is 81.8%. As for the CNN method, the highest accuracy obtained is 89% with a loss value of 0.0008 and the lowest accuracy is 75% and an average accuracy of 81.84%. At their highest accuracy, both methods require 6 seconds of practice time.

Item Description

http://repository.upnvj.ac.id/19858/1/ABSTRAK.pdf
http://repository.upnvj.ac.id/19858/2/AWAL.pdf
http://repository.upnvj.ac.id/19858/3/BAB%201.pdf
http://repository.upnvj.ac.id/19858/4/BAB%202.pdf
http://repository.upnvj.ac.id/19858/5/BAB%203.pdf
http://repository.upnvj.ac.id/19858/6/BAB%204.pdf
http://repository.upnvj.ac.id/19858/7/BAB%205.pdf
http://repository.upnvj.ac.id/19858/8/DAFTAR%20PUSTAKA.pdf
http://repository.upnvj.ac.id/19858/9/RIWAYAT%20HIDUP.pdf
http://repository.upnvj.ac.id/19858/10/LAMPIRAN.pdf
http://repository.upnvj.ac.id/19858/11/HASIL%20PLAGIARISME.pdf
http://repository.upnvj.ac.id/19858/12/ARTIKEL%20KI.pdf