Optimization Sentimen Analysis using CRISP-DM and Naive Bayes Methods Implemented on Social Media

Abstract

Freedom of expression on social media Twitter not always give positive value, because sometimes can contains negative things such as fake news, spreads hate speech, and racism, where these kinds of tweet can be categorized as an act of Cyberbullying. Where this cyberbullying tends to increase every time. The aim of this study is to use the Naïve Bayes method in classifying types of sentiment on Twitter. The keyword used is Saipul Jamil, and the tweet was taken in September 2021. A total of 18,067 tweets were collected and then they will be labelled with a positive or negative value. This study also uses the CRIPS-DM method which is consist of Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment stages. The results of this study obtained the value of Accuracy (85.6%), Negative Recall (82.1%), Positive Recall (90.23%), and Negative Precision (91.76%) Positive Precision (79.18%).