Offensive Language Detection on Social Media Based on Text Classification

Abstract

There is a concerning rise of offensive language on the content generated by the crowd over various social platforms. Such language might bully or hurt the feelings of an individual or a community. Recently, the research community has investigated and developed different supervised approaches and training datasets to detect or prevent offensive monologues or dialogues automatically. In this study, we propose a model for text classification consisting of modular cleaning phase and tokenizer, three embedding methods, and eight classifiers. Our experiments shows a promising result for detection of offensive language on our dataset obtained from Twitter. Considering hyperparameter optimization, three methods of AdaBoost, SVM and MLP had highest average of F1-score on popular embedding method of TF-IDF.