Sentiment Analysis of Police Performance On Twitter Users Using Naïve Bayes Method

Abstract

After the case of alleged child rape in East Luwu which was stopped went viral in the aftermath of many other cases of sexual violence that were considered by the police to be inconsistent with procedures. After the hashtag #PercumaLaorPolisi appeared #PolriSesuaiProsedur hashtag became a trending topic on Twitter. This study discusses the sentiment of police performance on twitter users, aiming to measure how much sentiment the performance of the police according to twitter citizens who earned. The topic of this study is a mining text that uses the naïve bayes method. Text mining is a computer-based algorithmic technique/approach to gaining new knowledge hidden from a set of texts. The data from crwalling on twitter were analyzed using naive bayes which is a method for analyzing. Naive Bayes' algorithm is very effective in classification or classification problems. This algorithm works based on existing probabilities to determine the probability of the future. The steps in the Naïve Bayes method are preprocessing which includes transformation, tokenization and filtering processes. It is followed by the weighting of words such as TF-IDF and ends with classificationand evaluation. As a result of this study, according to tweet data processed using the orange application and confusion matrix calculations, the police performance sentiment entered the neutral classification of 75.8%, negative 58.1% and positive 39.5% in the last order, as well as the resulting model at an accuracy value of 0.929, precision 0.933, recall 0.923, and f-measure0.954