Similarity : SENTIMENT ANALYSIS USING RECURRENT NEURAL NETWORK-LSTM IN BAHASA INDONESIA

The opinion of different individuals is an essential factor that needs to be
considered in decision making. Due to the current competitive business
environment, understanding customer opinions determine the success of modern
companies. Various machine learning algorithms are used in Automatics opinion
polarization mining from online text sources such as social media, user
comments, and reviews. This approach has been used in many languages,
including Indonesian. The purpose of this study therefore is to proposes the
Recurrent Neural Network (RNN)-Long Short Memory Term (LSTM) in
classifying sentiment polarity of Indonesian sentences. It evaluates the proposed
algorithm with a dataset from traveling site reviews consisting of 25,000 reports
in two classes of equal proportion (positive and negative). According to the
evaluation results, the model has 95.0% accuracy.