Abstract

Indonesian people convey opinions to public officials by involving community organizations in demonstrations.  However, due to digital era, many people also choose to respond the performance of public officials by conveying it through social media, one of which is Twitter. Sociaty opinion recorded on Twitter can be used for structured analysis using sentiment analysis. Sentiment analysis aims to shape data into specific classes. The class classification in sentiment analysis is in the form of positive classes and negative classes. This study applies the Naïve Bayes algorithm in classifying the sentiment of Twitter data, sociaty assessments of public officials. The data used came from text data of 8000 Tweets which was then preprocessed to produce 7993 data for sentiment analysis. Evaluation of algorithm performance using confusion matrix to obtain accuracy and error rate values. The results of sentiment analysis show that the assessment of people with the highest frequency is in the negative class. The algorithm performance shows an accuracy value of 64.55% with an error rate of 35.45%.