Unmasking Fake Users on Social Media: An AI-Driven Detection Framework

Öz

The widespread use of fake user accounts on social media platforms poses serious threats to digital trust, public opinion, and platform integrity. This paper presents a robust, AI-powered detection framework that integrates behavioral analytics, contextual text representation, and social graph structure learning. Our model combines a fine-tuned BERT (Bidirectional Encoder Representations from Transformers) model for textual analysis and a Graph Neural Network (GNN) built with PyTorch Geometric to capture network-level anomalies. Behavioral metadata such as account age, follower count, and posting frequency are also incorporated into the feature set. We employ SHAP (SHapley Additive exPlanations) for explainability, allowing detailed attribution of predictions to specific input features. The framework is evaluated using two public benchmark datasets: the Cresci-2017 dataset and a manually-labeled Twitter dataset, both of which include profile metadata, textual content, and interaction histories. All experiments were conducted on an Ubuntu 22.04 workstation with an NVIDIA RTX 3090 GPU and 64GB RAM. Our hybrid BERT+GNN model achieved state-of-the-art performance, with 94% accuracy and an F1-score of 0.93, significantly outperforming Random Forest, SVM, and single-modality deep learning baselines. We further analyze fake user behavior through heatmaps and word cloud visualizations. This study provides a scalable and explainable solution for detecting fake users, with potential applications in real-time moderation, bot detection, and information credibility assessment. Future work will focus on multimodal content integration (e.g., images, videos), real-time system deployment, and adaptive learning against evolving threat behaviors.

Anahtar Kelimeler

• Fake users
• social media analysis
• deep learning
• BERT
• graph neural networks
• explainable AI

Sosyal Medyada Sahte Kullanıcıları Ortaya Çıkarmak: Yapay Zeka Tabanlı Bir Tespit Çerçevesi

Öz

Sosyal medya platformlarında sahte kullanıcı hesaplarının yaygın olarak kullanılması, dijital güven, kamuoyu algısı ve platform bütünlüğü açısından ciddi tehditler oluşturmaktadır. Bu çalışma, davranışsal analiz, bağlamsal metin temsili ve sosyal grafik yapısı öğrenimini entegre eden sağlam bir yapay zeka tabanlı tespit çerçevesi sunmaktadır. Modelimiz, metinsel analiz için ince ayar yapılmış bir BERT (Bidirectional Encoder Representations from Transformers) modeli ile, ağ düzeyinde anormallikleri yakalamak amacıyla PyTorch Geometric kullanılarak geliştirilmiş bir Grafik Sinir Ağı (GNN) mimarisini birleştirmektedir. Hesap yaşı, takipçi sayısı ve paylaşım sıklığı gibi davranışsal metaveriler de özellik kümesine dahil edilmiştir. Açıklanabilirlik sağlamak adına SHAP (SHapley Additive exPlanations) yöntemi kullanılmış ve tahminlerin hangi girdilere dayandığı ayrıntılı olarak analiz edilmiştir. Çerçevemiz, profil metaverileri, metin içerikleri ve etkileşim geçmişlerini içeren iki kamuya açık veri kümesi üzerinde test edilmiştir: Cresci-2017 veri kümesi ve elle etiketlenmiş bir Twitter veri kümesi. Tüm deneyler, NVIDIA RTX 3090 GPU’ya ve 64GB RAM’e sahip bir Ubuntu 22.04 çalışma istasyonunda gerçekleştirilmiştir. Önerilen BERT+GNN hibrit modelimiz, %94 doğruluk ve 0.93 F1 skoru ile, Random Forest, SVM ve tek modelli derin öğrenme yöntemlerine kıyasla anlamlı şekilde daha yüksek performans göstermiştir. Ayrıca, sahte kullanıcı davranışları ısı haritaları ve kelime bulutu görselleştirmeleriyle analiz edilmiştir. Bu çalışma, gerçek zamanlı moderasyon, bot tespiti ve bilgi güvenilirliği değerlendirmeleri gibi uygulamalarda kullanılabilecek ölçeklenebilir ve açıklanabilir bir sahte kullanıcı tespit çözümü sunmaktadır. Gelecek çalışmalar, çok modelli içerik entegrasyonu (örneğin görseller, videolar), gerçek zamanlı sistem uygulamaları ve değişen tehdit davranışlarına karşı uyarlanabilir öğrenme yaklaşımlarına odaklanacaktır.

Anahtar Kelimeler

• Sahte kullanıcılar
• sosyal medya analizi
• derin öğrenme
• BERT
• grafik sinir ağları
• açıklanabilir yapay zeka

Kaynakça

1. Ahmed, F., & Abulaish, M. (2013). A generic statistical approach for spam detection in online social networks. Computer Communications, 36(10–11), 1120–1129. https://doi.org/10.1016/j.comcom.2013.04.004
2. Al-Qurishi, M., Alrubaian, M., Alomar, A., Alamri, A., Al-Rakhami, M., & Al-Khaleefa, A. (2021). Bot detection using content and link analysis in social networks. IEEE Access, 9, 101271–101284. https://doi.org/10.1109/ACCESS.2021.3096802
3. Badawy, A., Ferrara, E., & Lerman, K. (2018). Analyzing the digital traces of political manipulation: The 2016 Russian interference Twitter campaign. In Proceedings of the IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) (pp. 258–265). IEEE. https://doi.org/10.1109/ASONAM.2018.8508646
4. Cao, G., Li, H., Liu, Z., Yu, Y., & Wu, J. (2022). Fake account detection using co-training and active learning. Knowledge-Based Systems, 240, 108080. https://doi.org/10.1016/j.knosys.2021.108080
5. Chen, Y., Zhang, H., & Li, X. (2022). Bot detection using behavior tree analysis. In Proceedings of the 31st ACM International Conference on Information and Knowledge Management (CIKM) (pp. 3630–3634). https://doi.org/10.1145/3511808.3557729
6. Cresci, G., Di Pietro, R., Petrocchi, M., Spognardi, A., & Tesconi, M. (2015). Fame for sale: Efficient detection of fake Twitter followers. Decision Support Systems, 80, 56–71. https://doi.org/10.1016/j.dss.2015.09.003
7. Cresci, S., Lillo, F., Regoli, D., Tardelli, S., & Tesconi, M. (2020). The paradigm-shift of social spambots: Evidence, theories, and tools for the arms race. ACM Computing Surveys, 52(3), 1–34. https://doi.org/10.1145/3342559
8. Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) (pp. 4171–4186). https://doi.org/10.48550/arXiv.1810.04805

9. Ferrara, E. (2020). Disinformation and social bot operations in the COVID-19 era. Harvard Kennedy School Misinformation Review, 1(3), 1–8. https://doi.org/10.37016/mr-2020-042
10. Ferrara, E., Varol, O., Davis, C. A., Menczer, F., & Flammini, A. (2016). The rise of social bots. Communications of the ACM, 59(7), 96–104. https://doi.org/10.1145/2818717
11. Gao, H., Hu, J., Huang, T., Wang, J., & Chen, Y. (2018). Understanding social bot behavior using graph embedding. In Proceedings of the World Wide Web Conference (WWW) (pp. 1459–1468). https://doi.org/10.1145/3178876.3186066
12. Islam, R., Mahmud, M., & Al Mamun, A. (2021). Social media moderation with XAI: Challenges and solutions. In Proceedings of the IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) (pp. 432–435). https://doi.org/10.1109/ASONAM.2021.9519245
13. Jain, P., Kumar, P., & Chakraborty, S. (2022). Combining GNN and BERT for robust fake user detection. In Proceedings of the ACM Web Conference 2022 (pp. 361–370). https://doi.org/10.1145/3485447.3512179
14. Kipf, T. N., & Welling, M. (2017). Semi-supervised classification with graph convolutional networks. In Proceedings of the International Conference on Learning Representations (ICLR). https://doi.org/10.48550/arXiv.1609.02907
15. Kumar, S., & Carley, K. M. (2019). Tree LSTM with hierarchical attention for user profiling in social media. In Proceedings of the International AAAI Conference on Web and Social Media (ICWSM) (pp. 274–285). https://ojs.aaai.org/index.php/ICWSM/article/view/3222
16. Lin, H., Zhang, J., Chen, Y., & Luo, X. (2020). Detecting spammers on Twitter using learning automata. In Proceedings of the IEEE International Conference on Multimedia and Expo (ICME) (pp. 1–6). https://doi.org/10.1109/ICME46284.2020.9102911
17. Liu, Y., & Zhang, J. (2019). Fake user detection using graph convolutional networks. In Proceedings of the 2019 IEEE International Conference on Data Mining (ICDM) (pp. 1130–1135). IEEE. https://doi.org/10.1109/ICDM.2019.00136
18. Mihaylov, T., & Nakov, P. (2016). Hunting for troll comments in news community forums. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (ACL) (pp. 399–405). https://doi.org/10.18653/v1/P16-2065
19. Nguyen, H., & Le, T. (2022). Fake user detection on Instagram using neural attention mechanisms. IEEE Access, 10, 88243–88251. https://doi.org/10.1109/ACCESS.2022.3200764
20. Pan, S. J., & Yang, Q. (2010). A survey on transfer learning. IEEE Transactions on Knowledge and Data Engineering, 22(10), 1345–1359. https://doi.org/10.1109/TKDE.2009.191
21. Papernot, N., McDaniel, P., Wu, X., Jha, S., & Swami, A. (2016). The limitations of deep learning in adversarial settings. In Proceedings of the 2016 IEEE European Symposium on Security and Privacy (EuroS&P) (pp. 372–387). IEEE. https://doi.org/10.1109/EuroSP.2016.36
22. Pennington, J., Socher, R., & Manning, C. D. (2014). GloVe: Global vectors for word representation. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 1532–1543). https://doi.org/10.3115/v1/D14-1162
23. Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). “Why should I trust you?”: Explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD) (pp. 1135–1144). https://doi.org/10.1145/2939672.2939778
24. Shu, K., Wang, S., Lee, D., & Liu, H. (2020). Combating disinformation in a social media age. ACM SIGMOD Record, 49(3), 37–42. https://doi.org/10.1145/3424000.3424004
25. Tee, A. R., Chan, J. C. C., & Kwok, R. H. (2023). Real-time bot detection with XAI. IEEE Transactions on Neural Networks and Learning Systems, 34(5), 2456–2468. https://doi.org/10.1109/TNNLS.2021.3137967
26. Upadhyay, S., & Mishra, P. (2021). Explainable AI for social media bot detection. In Proceedings of the 2021 IEEE International Conference on Machine Learning and Applications (ICMLA) (pp. 345–352). IEEE. https://doi.org/10.1109/ICMLA52953.2021.00061
27. Varol, O., Ferrara, E., Davis, C. A., Menczer, F., & Flammini, A. (2017). Online human-bot interactions: Detection, estimation, and characterization. In Proceedings of the 11th International AAAI Conference on Web and Social Media (ICWSM) (pp. 280–289). https://ojs.aaai.org/index.php/ICWSM/article/view/14972
28. Wang, H., Zhang, F., Hou, M., Xie, X., & Guo, M. (2020). Detecting fake accounts on social networks using semi-supervised learning. IEEE Access, 8, 57612–57620. https://doi.org/10.1109/ACCESS.2020.2981706
29. Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., & Philip, S. Y. (2021). A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 32(1), 4–24. https://doi.org/10.1109/TNNLS.2020.2978386
30. Yu, B., Chen, Z., Wang, L., & Wang, S. (2022). Temporal patterns in bot behavior on Twitter. In Proceedings of the ACM Web Conference 2022 (WWW) (pp. 1741–1750). https://doi.org/10.1145/3485447.3511992
31. Zhang, J., & Paxson, S. (2011). Detecting and analyzing automated activity on Twitter. In Proceedings of the International Conference on Passive and Active Network Measurement (PAM) (pp. 102–111). Springer. https://doi.org/10.1007/978-3-642-19260-9_11
32. Zhang, X., & Zhao, L. (2022). BERT meets GCN: A hybrid approach for detecting fake users. In Proceedings of the 2022 IEEE International Conference on Big Data (Big Data) (pp. 2107–2116). IEEE. https://doi.org/10.1109/BigData55660.2022.10020569