Multi-Label Classification of Indonesian Hate Speech Detection Using One-vs-All Method
Ema Utami, Rini Rini, Ahmad Fikri Iskandar, Suwanto Raharjo · 2021
Hate speech is an act of communication carried out against specific targets, categories, and levels related to abusive language, even though abusive language is not necessarily. Currently, text mining technology is one approach that can overcome hate speech. Conventional Hate speech expressed in Indonesian Language on Twitter tranform single label to multi-label. The novelty of this research to classifying multi-label hate speech in Indonesian Language using One-vsAll (OvA) that testing on several models of machine learning. Multi-Label on hate speech using in this research is Abusive, HS_Weak, HS_Moderate, and HS_Strong, then transform into new label OvA. There are two scenarios of feature extraction: Bag of Word (BoW) and Term Frequency-Inverse Document Frequency (TF-IDF). The experimental results showed that the Artificial Neural Network (ANN) classifier with the One-vs-All method, BoW and Chi-square feature selection got the best accuracy of 86.96%