Detection of Fake Social Media Accounts Using Machine Learning

Fake social media accounts may be used for spam, artificial follower growth, impersonation, coordinated activity, and manipulation of online discussions. Their detection is challenging because deceptive accounts can imitate normal users, while genuine users may also show unusual activity. This paper presents a literature-based review and a conceptual machine-learning framework for account-level screening. The framework combines profile, activity, content, and network signals, applies preprocessing and feature engineering, and compares supervised classifiers such as Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, and Naïve Bayes. Explainability techniques, including SHAP or LIME, may help reviewers understand the factors behind a suspicious prediction. The proposed workflow also defines an evaluation plan based on precision, recall, F1-score, and confusion-matrix analysis. No trained model, experimental dataset, or measured performance result is claimed in this paper. The contribution is a structured conceptual design that highlights task definition, evaluation discipline, explainability, privacy, and human review as requirements for a responsible account-detection system