Sentiment Analysis of Indonesian Netizen Opinion about Covid-19 Vaccine

Abstract

In 2021, Indonesia government performed National Covid-19 vaccine program throughput Indonesia. Al though it was considered a successful program, based on the number of participation, there were also quite many negative opinion about Covid-19 vaccine, especially online. The purpose of this research is to crate a sentiment analysis on Indonesian opnion about Covid-19 vaccine based on their tweets on Twitter. We collected 4000 tweets, but after bing preprocessed and oversampling, we gat a dataset consisted of 1685 balanced data of two classes, Positive opinion and Negative opinion. The classifier model was built with Multi Layer perceptron Neural Network algorithm. The model has 68,8% accuracy with 0,82 precision, and 0,62 recall. The low result on recall was caused by nonoptimal feature extraction, especially for the Negative class features. There were few features in Negative class that were also a feature in Positive class. Other that that, the high use of abbreviations and non-standard words were also the hindrance for the feature extraction process which are stopwords removal and stemming could perform optimal.