AUTOMATIC DETECTION OF ABUSE ON SOCIAL MEDIA

Social media over the few of its existence has been a technological innovation, which has made information communications easily accessible to the all groups of people from all walks of life. But, just as every great tool has its drawbacks, social media also has its own drawbacks and risks connected with it. Social media has led to the increase in cases of cyberbullying whereby a group of people or individuals are intimidated or threatened with messages from other people. Every user of social media is vulnerable to abuse on social media. The main objective of this research is to study abuse on social media and to develop a real-time system that will able to automatically detect these threats and abuses. This document focuses on the Twitter social networking service as a case study. The study used a qualitative methodology to acquire an understanding of social media, the risks associated with and how to detect abuse. A model for a real-time analytical system which would receive a stream of Twitter data and analyze the level of abuse was documented. The methodology involved a proposition of the use of cloud computing technology with the aim of keeping the real-time processes running uninterrupted. The Microsoft Azure platform is suitable. By interfacing the RStudio IDE (Integrated Development Environment) with the Twitter API (Application Programming Interface), the R language was used to analyze the twitter stream data stored in a dataset and generated various real time reports of abusive words based on our set preferences and data mining techniques.