Improving Random Forest Method to Detect Hatespeech and Offensive Word
Kristiawan Nugroho, Edy Noersasongko, Purwanto Purwanto, Muljono Muljono, Ahmad Zainul Fanani, Affandy, Ruri Suko Basuki · 2019 International Conference on Information and Communications Technology (ICOIACT) · 2019
Hate Speech is a problem that often occurs when someone communicates with each other using social media on the Internet. Research on hate speech is generally done by exploring datasets in the form of text comments on social media such as Twitter, Facebook and MySpace. This study aims to improve the performance of the Random Forest method in detecting hatespeech and crude speech. In this paper the researcher uses a twitter hate speech and offensive identification dataset that is classified using the Random Forest method which will be compared with the results of its accuracy with AdaBoost and Neural Network to detect hatespeech and crude speech. The detection results of hatespeech and crude speech identification resulted in an accuracy of 0.722 for the Random Forest method and 0.708 using AdaBoost and 0.596 using Neural Network method.