Hate Speech Detection in Indonesian Language on Instagram Comment Section Using Maximum Entropy Classification Method

Social media nowadays is a platform for many things and has become a place to delivers opinions. Opinions in a form of hate speech are one of the problems that authorities find hard to solve, because of its number and variations. Instagram, with almost 100 million users in Indonesia, is the object of this research because it is one of the most popular social media in Indonesia where people can share their photo and video, has become one of the most popular places to express hatred towards others. Because of that, a system will be made to detect hate speech on Instagram using Maximum Entropy classification algorithm. TF-IDF is also used as the feature extraction method, with the improvement from Part of Speech (POS) Tagging as the parameter. Accuracy generated from this research is 86,67% using Maximum Entropy with POS parameter TF-IDF and 80% when using TF-IDF without POS parameter. This system is expected to determine whether a comment on Instagram is a hate speech or not, and to see how TF IDF and POS Tagging improve the performance of Maximum Entropy Classifier.