The program to carry out the COVID-19 vaccination has become an obligation
for the people of Indonesia. This affects the Indonesian people who provide sentiment
for the vaccine program in Indonesia. Indonesians usually comment via social media.
Commonly used social media are Twitter and Facebook. This study focuses more on
Twitter social media. The number of sentiment data obtained in the study was 16,291
tweets using 2 keys, namely "vaccinationcovid19" and "vaksincovid19". The data was
then categorized based on the calculatory value using lexicon resulting in positive
sentiment of 63.7%, negative sentiment of 26.2% and neutral sentiment of 10.1%. The
Naïve Bayes classification used is the Naïve Bayes Multinomial text classification and
Genetic Algorithm with the Feature Selection function. Before classifying the data
carried out Weighting with TF-IDF, Smote For Ibalance data and tweet data are
divided into 2 parts, namely testing data of 20% and training data of 80%. The
classification results using Naïve Bayes Multinomial accuracy of 82% and GA feature
selection optimization results in 82% accuracy. In this study, the discussion and
findings carried out by the Naïve Bayes test were optimized with GA featur selection
resulting in the most optimum accuracy of 82%.