Sentimen Twitter terhadap PILKADA kota Medan menggunakan metode Naive Bayes

Prasetyo Mimboro

JNANALOKA
Vol. 03 No. 01 Maret Tahun 2022

Indonesia is the fifth largest country with twitter users with 19.5 million users. Along with the development of information technology, twitter has become a source of information based on twitter sentiment and trending as well as the use of hashtags that are trending. Recently, the archipelago vaccine has reaped the pros and cons, to be able to classify positive and negative sentences in twitter sentiment towards the archipelago vaccine, it requires data from twitter users by taking data based on sentence classification which is then processed in the initial data before being entered into the indoBERT model which will later be resulting in the accuracy of twitter sentiment towards the archipelago vaccine. Indonesia has 19.5 million Twitter users out of a total of 500 million global users and continues to grow from time to time. Twitter users used it as an open forum for campaigns by the Medan mayoral candidate and their volunteers were asked by Netizens to respond. Netizens' responses to each tweet are both Positive and Negative. Therefore, this study tries to analyze tweets about netizens' sentiments towards the 2020 Medan City Election. Opinions or sentiments from Twitter users can of course be used as criticisms and suggestions that can be accommodated by candidates for mayor and deputy mayor of Medan. Twitter netizens often have opinions about Regional Head Candidates through their uploads. The opinions of Twitter Netizens are still random or unclassified. To facilitate the process of classifying netizen opinion data requires Sentiment Analysis. Sentiment analysis was carried out by classifying tweets containing Netizen sentiments towards the 2020 Medan City Election. The classification method used in this study is the Naive Bayes method combined with TF-IDF feature extraction. NS The validity test applied in this study used a confusion matrix. With the tf-idf extraction feature and the Naive Bayes method, it will be able to automatically classify sentiment analysis with an accuracy of 76.00%.