Analysis of Twitter Sentiment Towards Madrasahs Using Classification Methods

Abstract

In today's digital era, the influence and use of the internet has become a necessity, especially in Indonesia itself, internet users in early 2021 reached 202.6 million people. The most widely used internet use by Indonesians is social media. Several incidents of sexual violence that occurred in the madrasa environment as reported in the media, the emergence of radical Islamic issues which he said were the fruit of thoughts from the madrasa environment, terrorism which was also said to come from misinterpreting knowledge from madrasahs, intolerance to different religions, changes in the character of madrasah students and so on will cause negative thoughts towards  madrasah. To find out how the sentiment of social media users towards madrasahs, a study was conducted on analisis twitter sentiment towards madrasah using the classification method. The methods used are Naïve Bayes (NB), Decision Tree (DT) and K – Nearest Neighbor (K-NN).   Toimprove the performance of the classification method is carried out using the Particle Swarm Optimization (PSO) selection feature.   On the other hand, tools gataframework, execute Python script dan rapidminer diguna kan jug a dalam penelitian this to membantu preprocessi ng dan cleansing pa da datasethingga membantu menciptaka n corpus dan sentiment ana lysis.   Acuration obtained from the Naïve Bayes algorithm accuracy: 76.86% +/- 5.24% (micro average: 76.86%), Decision Tree accuracy: 61.38% +/- 5.46% (micro average: 61.35%), K-NN accuracy: 74.70% +/- 4.83% (micro average: 74.67%), Naïve Bayes PSO accuracy: 80.80% +/- 4.86% (micro average: 80.79%, Decision Tree PSO accuracy: 65.27% +/- 5.26% (micro average: 65.28%), and K-NN PSO accuracy: 67.24% +/- 7.92% (micro average: 67.25%).  The results showed that the Naïve Bayes PSO algorithm got the best and accurate results. This study succeeded in obtaining an effective and best algorithm in classifying positive comments and negative comments related to sentiment analysis towards madrasahs by classification method.