Social Bot Detection using Machine Learning Algorithms: A Survey and Research Challenges

Abstract

In the past decade social media platforms growing rapidly and they are part of our routine
life. Each platform has its own specification which uses for specific purposes. After this
widely spread, those SMPs were targeted by the cybercriminals to cast their malicious
activities. There are many different malicious activities in SMPs such as spamming,
phishing, fake account. In these papers, Bots activities in SMPs one of those threats which
include fake accounts, fake friends/followers, spreading misinformation by purpose, and
many more. At the beginning of our work, we explain all terminology related to this topic to
have a clear understanding of what is going on now. Then we reviewed the recent papers
about this topic. We found out different models suggested by the researchers for recognizing
those malicious activities. Until now most of the work focusing on Twitter as a platform,
English as a language, and machine learning as a detection method but there are many gaps
in this research area because Twitter is the 17th most used SMPs in 2020, also there are
many malicious actions in other languages, and detection method needs lots of improvement
in reliability, accuracy, real-time detection, and performance area. As a result, we are at the
beginning of the game and we need lots of improvement for controlling the bot’s activities.
Besides all technical term also people awareness has a big impact on controlling a bot
because most of the times the botmaster use people ignorance to make their actions easy.