Sentiment Analysis on Hate Speech Post 2024 Election for Elected President Using a Hybrid Model Machine Learning

One of the important events in the democratic life of a country is the general election. In addition, the possibility of hate speech appearing on social media increases as political tensions increase. This hate speech can take the form of negative comments, insults, or even threats against the elected president. This research uses the content of tweets as a data source to analyze public opinion and sentiment towards the elected president. This research aims to analyze sentiment towards hate speech held by twitter users towards the elected president after the 2024 election by building a hybrid model using 3 algorithms, namely k-nearest neighbors, long short-term memory and naive bayes. The results of tests carried out with 12,000 tweet data that show the naive bayes method classification results have an accuracy of 72%, the long short-term memory classification results show an accuracy of 78%, the k-nearest neighbors method accuracy value is 83%, and the hybrid model accuracy value is 78%. Considering the accuracy values of the three algorithm method, by using a hybrid model we can increase the accuracy by combining the three algorithm models. from previously having the lowest accuracy of 72%, by using a hybrid model we can increase the accuracy to 78%.