Intentions of Twitter Automation: A Study of Tweet Sources

Abstract

As a global real-time communications platform with over 400 million monthly visitors around the world, Twitter is considered to be one of the most influential social networks. People of all age groups use Twitter to daily consume and create data in the form of tweets. Twitter can hence be used to reach and affect a wide audience. Thus, it is important to regulate the accounts in Twitter to detect and prevent any malicious campaign. Previously, Twitter has been used to spread spam and malware and to collect private information of Twitter users. Previous work concerned with securing Twitter via detecting such accounts is divided into two groups. Their common aim is to design and implement algorithms that identify potentially harmful twitter accounts. The first group focused on detection of automated accounts controlled by computer software. Detecting automation alone isn’t sufficient however as an automated account isn’t necessarily harmful. On the contrary, there are many Twitter bots that are benign and even useful to the Twitter community such as bot accounts that automatically alert people of natural disasters. The second group focused on detecting accounts with a certain malicious intention such as spam or political astroturf. This approach has a limitation of only detecting the accounts with that particular malicious intention targeted. It is not comprehensive as there are several different kinds of malicious Twitter accounts. In this work, we aim to overcome the limitations of both approaches by creating a tool that combines automation detection with intent identification while introducing improvements in the two areas. Those improvements are built upon analysis and classification of tweet sources. We designed a classification system for tweet sources that we then used to detect automation and identify intentions of Twitter accounts.

Date of Award	2018
Original language	American English
Awarding Institution	HBKU College of Science and Engineering