Toxic Comments Identification in Arabic Social Media

Abstract

The usage of social media increases day by day. Both individuals and organizations use social media for different purposes. Problems increase in association with social media technologies. Toxic comments bots create a negative impression about people, companies and products. These kinds of toxic comment bots are created by the attackers. This research work is carried out to identify these toxic comments in Arabic social media. For that, Machine learning techniques are used. Mainly gradient boosting technique (XGBoost algorithm) has been utilized to effectively identify the comments created by the toxic comment bots. XGBoost can efficiently divide toxic comments into the following categories, toxic, severe toxic, obscene, threat, insult, and identity hate. The accuracy achieved by the proposed method is 99.54 %.