International Journal of Scientific Research and Engineering Development

Social media platforms have seen a sharp rise in fake, automated, and impersonated accounts, making it harder to preserve trust, control misinformation, and maintain platform safety. This paper proposes a quantum-inspired machine learning framework for evaluating whether a social media account is authentic by learning useful patterns from commonly available profile, content, and interaction features. These include signals such as account age, posting frequency, timing regularity, follower-following relationships, engagement behaviour, writing-style indicators, and network connectivity. To study performance, multiple classifiers including logistic regression, decision tree, random forest, and gradient boosting are trained and tested using stratified train-test splits. The models are compared using metrics such as accuracy, precision, recall, F1-score, and ROC-AUC, with particular focus on minimizing false positives for genuine users and false negatives for abusive accounts. The best model is then deployed through a user-friendly web application that provides probability-based authenticity predictions along with confidence scores and interpretable explanations supported by SHAP visualizations. These explanations help moderators and analysts understand which features influenced each prediction, improving transparency and supporting better decision-making. Overall, the work shows that combining quantum-inspired learning with interpretable machine learning can strengthen fake-account detection, improve platform reliability, and support scalable authenticity assessment in modern social media systems.

Namana S, Monisha B S, Rachana T M, Vanishree B S,"Quantum-Inspired Machine Learning Framework for Social Media Account Authentication" International Journal of Scientific Research and Engineering Development, V9(4): Page(53-60) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.