<![CDATA[The presidential election in Indonesia is a frequently discussed topic on social media, especially Twitter. This platform provides a space for people to express their views on presidential candidates and election issues, making it suitable as a data source for this study. This study aims to analyze public sentiment towards presidential election news on Twitter using the Naïve Bayes Classifier method. Data was taken from Twitter for the period 5–13 February 2024 with a total of 2,561 comments. The research process includes data collection, preprocessing, data labeling, and model training and testing. Naïve Bayes was chosen because it is efficient in text classification and has several variants for model experiments. Sentiment is classified into three main categories, namely positive, negative, and neutral. The results showed that negative comments dominated (41%), followed by positive (37.3%) and neutral (21.7%). The Multi Naïve Bayes Classifier model provided the highest accuracy (81%), followed by Bernoulli Naïve Bayes (80%) and Gaussian Naïve Bayes (76%). This difference in accuracy is influenced by the model's sensitivity to data characteristics, such as the number of features and sentiment distribution. This research has the potential to help campaign teams understand the issues that trigger negative responses and support policy makers in designing more effective political communication strategies.]]>
Copyrights © 2024
