RECURRENT NEURAL NETWORK (RNN) DENGAN LONG SHORT TERM MEMORY (LSTM) UNTUK ANALISIS SENTIMEN DATA INSTAGRAM

Rudy Cahyadi, Ariesta Damayanti, Dede Aryadani

Abstract


Social media is a new public space for channeling opinions and ideas. Social media such as Instagram has been used by STMIK AKAKOM Yogyakarta to provide information related to these high school institutions. This information can later become a topic of conversation and interesting things for the community to discuss. The response or public response to the Instagram content of STMIK AKAKOM Yogyakarta is of course very diverse. Therefore, the researcher tries to analyze the comments that talk about the content of the Instagram STMIK AKAKOM Yogyakarta.

Sentiment analysis was performed using the Recurrent Neural Network (RNN) method with Long Short Term Memory (LSTM). Comments will be identified whether the comments have positive, neutral or negative sentiment. This study uses data as much as 1,473 data obtained from the results of crawling.

The result of this research is a system that can classify sentiments. The level of testing accuracy obtained was 65% and the application accuracy rate obtained was 79.46%. Some of the obstacles in the process of sentiment analysis are the imbalanced dataset, so it is necessary to take additional steps in preparing the data for model training.

 


Keywords


deep learning, instagram, RNN, LSTM, sentiment analysis

Full Text:

PDF


DOI: http://dx.doi.org/10.26798/jiko.v5i1.407

Article Metrics

Abstract view : 4269 times
PDF - 2199 times

Refbacks

  • There are currently no refbacks.




Copyright (c) 2020 Rudy Cahyadi, Ariesta Damayanti, Dede Aryadani


JIKO (Jurnal Informatika dan Komputer)

Published by
Lembaga Penelitian dan Pengabdian Masyarakat
Universitas Teknologi Digital Indonesia (d.h STMIK AKAKOM)

Jl. Raya Janti (Majapahit) No. 143 Yogyakarta, 55198
Telp. (0274)486664

Website : https://www.utdi.ac.id/

e-ISSN : 2477-3964 
p-ISSN : 2477-4413