Hate speech content detection system on Twitter using K-nearest neighbor method

Twitter is a social media platform that many Indonesians use to express their thoughts on a variety of topics. In Indonesia, the use of social media is governed by a law known as Information and Electronic Transactions Law. However, until now, the implementation of this law has been subpar. This is because there are still violations occurring, and no legal action has been taken against these violations. Hate speech is a common violation on Twitter. The goal of this research is to create a system that can detect potential violations of content on Twitter, particularly content containing hate speech. The k-nearest neighbor (KNN) method was used in this research, along with the feature extraction method TF-IDF. The system built will detect whether the tweet you want to post violates a specific article in the Information and Electronic Transactions Law. Based on model validation, model classifier built has accuracy value is 67.86%, with K value in the KNN method is 10. Meanwhile, based on user validation, the system created has an accuracy of 77%.