ANALYSIS OF TWITTER SENTIMENT TOWARDS MADRASAHS USING CLASSIFICATION METHODS

Abstract

Several incidents of sexual violence, the emergence of radical Islamic issues, terrorism, intolerance of changes in the character of students and so on have recently become a highlight for madrasahs. To find out how the sentiment of social media users towards madrasahs, research on twitter sentiment towards madrasahs was conducted using text mining techniques. The methods used are Naïve Bayes (NB), Decision Tree (DT) and K – Nearest Neighbor (K-NN) which aim to classify public sentiment towards Madrasahs on Twitter. The dataset used is a tweet in Indonesian with the keyword "Madrasah" as many as 3288 tweets. The techniques used to build classification and sentiment analysis are text mining, transformation, tokenize, stemming and classification, etc. Gataframework tools, execute Python script and RapidMiner are also used to help create sentiment analysis in measuring classification values. The results obtained by the optimization using Particle Swam Optimization (PSO) using the Naïve Bayes algorithm and the accuracy value obtained was 80.80%, with a precision value of 83.03%, a recall value of 78.68%, and an AUC of 0.739.