Sentiment analysis is a field of study that analyzes one's opinions, sentiments, evaluations, attitudes and emotions that are conveyed in written text. There are several factors that cause low accuracy results from sentiment analysis. These factors such as less optimal stemming process, word negation process that does not produce maximum results, writing errors in the dataset, and others. These problems can be overcome by optimizing the process of normalizing words, negation, stemming, and adding methods of semantic expansion. The purpose of adding the Semantic Expansion method and improvement in the process is to increase the accuracy value of the Sentiment Analysis process. This study aims to create a sentiment analysis model from public comments on a public figure (Ridwan Kamil) using the Naïve Bayes Classifier algorithm. Based on the test results in the sentiment analysis model using the Naïve Bayes Classifier method with the addition of the semantic expansion method it is proven that it can improve accuracy. The accuracy obtained using the semantic expansion method is 72%. While the value of accuracy without semantic expansion is 70%.