EUVIMEDEUVIMED entdecken
Crossref

An Integrated Machine Learning Framework for Spammer and Fake User Detection in Online Social Networks

Mounika Kurme, Ramana Reddy N

Journals Fringe Multi-Engineering Proceedings · 2025

DOI: 10.69996/fmep.2025015 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journals Fringe Multi-Engineering Proceedings, 2025. Verfasst von Mounika Kurme und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper proposes a robust machine learning-based framework for detecting spammers and fake users within Online Social Networks (OSNs), addressing the critical challenges of security, privacy, and authenticity in modern social platforms. By analyzing a rich set of behavioral and profile features—such as account age, posting frequency, follower-to-following ratio, and content characteristics—the system leverages ensemble classifiers like Random Forest to accurately distinguish between genuine and fraudulent accounts. The framework integrates explainable AI techniques to provide transparency and interpretability of detection decisions, while privacy-preserving mechanisms ensure compliance with data protection regulations. Extensive simulations demonstrate the model’s high accuracy, precision, recall, and low error rates, outperforming traditional classifiers and adapting effectively to evolving threats. This research offers a scalable, real-time, and trustworthy solution that enhances the security and user experience of OSNs, contributing significantly to ongoing efforts in combating fake accounts and spam on social media platforms. The proposed system analyzes key features such as account age, posting frequency, follower-to-following ratio, and content patterns, employing ensemble methods like Random Forest for classification. Simulation results demonstrate that the Random Forest model achieves an accuracy of 96.8%, precision of 95.7%, recall of 96.2%, and an F1-score of 95.9%, outperforming other classifiers such as Decision Tree and SVM. The system also maintains low false positive and false negative rates of 2.1% and 1.8%, respectively. Additionally, explainable AI techniques are integrated to provide transparency in decision-making, while privacy-preserving methods ensure compliance with data protection standards.

Zitieren

APA

Kurme, M., & N, R. R. (2025). An Integrated Machine Learning Framework for Spammer and Fake User Detection in Online Social Networks. Journals Fringe Multi-Engineering Proceedings. https://doi.org/10.69996/fmep.2025015

Vancouver

Kurme M, N RR. An Integrated Machine Learning Framework for Spammer and Fake User Detection in Online Social Networks. Journals Fringe Multi-Engineering Proceedings. 2025. doi:10.69996/fmep.2025015

BibTeX
@article{Kurme2025,
  author = {Kurme, Mounika and N, Ramana Reddy},
  title = {An Integrated Machine Learning Framework for Spammer and Fake User Detection in Online Social Networks},
  journal = {Journals Fringe Multi-Engineering Proceedings},
  year = {2025},
  doi = {10.69996/fmep.2025015}
}
Kostenlos starten →