SENTIMENT ANALYSIS BASED ON OPINION REVIEWS OF GOJEK USING THE LONG SHORT-TERM MEMORY (LSTM) METHOD

Abstract

Many people are very interested in using online services like GoJek, mainly when using public transportation or traffic jams is challenging. Based on assessments of online transport services, particularly GoJek, we suggest a technique in this study that can identify public attitudes. The 20,000 and 2,500 reviews from motorcycle taxis utilized as the dataset were divided into Positive, Neutral, and Negative categories. To complete this investigation, the LSTM model was used. The LSTM calculation model produced an accuracy value of 0.58% with a dataset of 2,535 positive, negative, and neutral label data; the best accuracy value for positive labels was 75.2%. For the 20,000-data collection, a significantly superior accuracy value of 0.87%, with a positive brand of 91.75%.