Manvir Dhinse Undergraduate Dissertation 2017/18

Identifying Socially Abusive Profiles on Twitter

Supervised by F.Ciravegna

Abstract

Twitter is amongst the most popular social media platforms; the microblogging platform receives over 500 million tweets per day and has over 328 million monthly active users. However, the nature of the platform and the ease of being able to create accounts with minimal details required, gives rise to the advent of Twitter Trolls. These profiles are primarily used for polluting the platform with socially abusive tweets on controversial and popular topics. This project aims to use relevant sentiment analysis and topic modelling techniques to create an automatable system to identify these types of profiles, capitalising on the common characteristics these profiles portray, which can then be further developed to maintain the number of Twitter Trolls on the platform.