Abusive Tweets Detection Using Supervised Learning With Contextual Features

Abusive Tweets Detection Using Supervised Learning With Contextual Features by Kamal Hussain (2020)

Social media are one of the most drastically growing platforms on the web which share and generate content in real-time. Among a diverse set of social media platforms, Twitter is the most widely used microblogging platform used to share millions of statuses every minute. Twitter enables people to publicly share short messages called tweets. Occasionally, some people use abuse in their tweets to offend others. It has severe consequences for public in general and targeted victims in particular. Many approaches have been proposed to detect abuse on Twitter which use content and lexicon based features. In this thesis, we propose an approach which uses contextual features including time window and sliding window. We use these features in various supervised machine learning algorithms and perform a diverse set of experiments, including various combinations of features used with a variety of supervised machine learning algorithms. We compare the results of the proposed approach with state-of-the-art methods. The results show that the proposed approach outperforms the state-of-art methods.