Identification of Hate Speech in Social Media

An exploration of different approaches to detect hate speech in social media is present in this paper. Due to the rapid growing of online content hate speech has become a common issue which can influence variety of hate crimes. So, there is a need to find an accurate and efficient technique to detect online hate content and flag them automatically. The experiment was carried out using a local English text dataset. Hate speech is defined as the usage of language to insult or spread hatred towards a group or individual based on religion, race, gender or social status for the experiment. Then a comparison of both supervised and unsupervised learning techniques with different feature types for the task of hate speech detection was done. From all the supervised and unsupervised models Naive Bayes classifier with Tf-idf features performed best with an F-score of 0.719.