Klasifikasi Lexicon-Based Sentiment Analysis Tragedi Kanjuruhan pada Twitter Menggunakan Algoritma Convolutional Neural Network

Abstract

This study aims to conduct a sentiment analysis of conversations on social media Twitter related to the Kanjuruhan Tragedy. Social media, especially Twitter, has become a significant platform for Indonesians to share their thoughts and feelings regarding this tragic event. We used two approaches for sentiment analysis, namely Lexicon-based and Convolutional Neural Network (CNN), with a focus on classifying sentiments in positive, negative, and neutral categories. This study also involves references to several previous studies that implemented various sentiment analysis methods. It is hoped that the results of this study can provide deep insight into the responses and feelings of the public on social media related to the Kanjuruhan Tragedy. The lexicon-based sentiment analysis classification of the Kanjuruhan Tragedy on twitter social media using the CNN algorithm successfully analyzed the sentiment results of tweets related to the tragedy where most of the tweets obtained had negative sentiments with test results of precision value 87.74%, recall 87.51%, and f1-score 87.27% with a classification accuracy of 87.27% and took 3 minutes 23 seconds of training time.