Hate speech and abusive language detection in Indonesian social media: Progress and challenges

Abstract

Nowadays Hate Speech and Abusive Language (HSAL) have spread extensively over social media. The easy use of social media allows people to abuse the media to spread HSAL. Hate speech and abusive language in social media must be detected because they can trigger conflict among citizens. Not only in social media, but HSAL also often trigger conflict in real life. In recent years, many scholars have researched HSAL detection in various languages and media. However, there are still many tasks on HSAL detection that need to be done to develop a better HSAL detection system. This paper discusses a summary of Indonesian HSAL detection research, conducted by utilizing the Kitchenham systematic literature review method. Based on our summary, we found that most Indonesian HSAL research still uses the classic machine-learning approach with classic text representation features that experimented on the Twitter text dataset. We also found several challenges and tasks that need to be addressed to build a better HSAL detection system in Indonesian social media that can detect the hate speech target, category, and levels; and the hate speech buzzer, thread starter, and fake account spreader.

Keywords: Abusive language; Hate speech; Indonesian social media.