Hate and Offensive Speech Detection Using Term Frequency-Inverse Document Frequency (TF-IDF) and Majority Voting Ensemble Machine Learning Algorithms

Abstract

The advancement in technology especially the internet has opened new frontiers to criminality and abuses of information. Social media have given racists and extremists a platform for carrying out their criminalities and attacks on legitimate users’ information. Thus, there is need to give adequate attention to the communications on social media so as to curtail these malicious acts before they materialize into causing physical harms. Hate speeches has been blamed for various degrees of violence experienced in the real world. A lot of research efforts have been put in detecting hate speeches using various techniques with varying degrees of accuracy and F-Measure. In this work, Term Frequency-Inverse Document Frequency (TF-IDF) with a majority voting ensemble learning classification Model were used for the detection of hate speech and a performance of 95% accuracy and 0.95 F-Measure were recorded