Analyzing User Profiles for Bot Account Detection on Twitter via Machine Learning Approach

Today social media platforms have become immensely popular. It is one of the ways in which we stay connected with our family and friends, thus making it an essential part of our lives. As the popularity of social media platforms increases, more and more users create their accounts and become a part of them. However, this popularity has also attracted many impersonators who create fake profiles and automated programs known as bots. These bots are involved in various malicious activities like directing users to spam websites, spreading false information about a person or organization, causing economic loss to society, and threatening the security and privacy of users. Such bots must be detected and removed. Thus, there is a need to design a system that can automatically analyze and detect such bots on social media platforms. In this paper, we have designed a system that can classify the bot and human accounts based on the user profile features. Machine learning algorithms like decision trees (DT), support vector machine (SVM), logistic regression (Log R), random forest, and neural networks are used for the classification task. Since the user profile features are huge, feature engineering and dimension reduction techniques like PCA are used. Inspired by the literature, new features are derived from existing features and used in the study. Finally, the comparative performance measure of accuracy, recall, precision, F1 score, and false positive rate of all the classifiers is presented.