Machine Learning Classifiers for Efficient Spammers Detection in Twitter OSN

Abstract

Online Social Networking (OSN) is one of the fastest-growing areas of communication over the internet. The popular social networking sites Facebook, LinkedIn, Instagram, Twitter etc. have attracted public helping them connected to family, friends and relatives. People share videos, texts, pictures, opinions, ideas, knowledge and some confidential information knowingly or unknowingly through these sites and thus, OSNs have become the main source of targets for cyber attackers. Cyberattacks have been increasing for the last few decades throwing a serious threat to the internet world. This paper first discusses various OSN threats such as misuse of identity, malware, phishing attacks etc. and also recommends some of the threats preventive measures. Since OSNs activities occur online, it is difficult to trust the users at the other end. Some users misguide genuine users (non-spammers). The one who misguides is known as spammers. Hence, our research work focuses on spammers detection in Twitter using machine learning models. Twitter data is used to carry out this research. For detection, three machine learning models are used, namely, Naïve Bayes, Support Vector Machine and Random Forest. Performance of the models is evaluated using Precision, Recall and F measure metrics.

Keywords: Online Social Networking, Security Threats, Cyber Security, Cyber Crime, Machine Learning, Support Vector Machine, Naïve Bayes, Random Forest