SEBOT-GCL: Graph Contrastive Learning for Social Bot Detection

Abstract

Social media platforms have experienced a rapid increase in automated accounts known as social bots, which are capable of spreading misinformation, spam, and malicious content. Detecting these bots is essential to maintain the reliability and security of online social networks. This paper proposes an intelligent social bot detection framework that utilizes graph-based learning and contrastive learning techniques to accurately identify automated accounts. The system constructs a social interaction graph where each node represents a user and edges represent interactions between users. A contrastive learning mechanism is used to learn meaningful representations of user behavior from multiple graph views. The proposed model analyzes structural and behavioral patterns to distinguish genuine users from bot accounts. Experimental evaluation performed on benchmark social media datasets demonstrates high accuracy and improved detection performance compared to traditional machine learning models. A user-friendly interface is also developed to visualize datasets, graph structures, and prediction results, allowing researchers and administrators to monitor suspicious activities effectively. The proposed framework provides a reliable and scalable solution for detecting social bots in large-scale social media networks.

Introduction

The text discusses the growing problem of social bots on platforms like Twitter, Facebook, and Instagram, where automated accounts mimic human behavior to spread spam, misinformation, and manipulate public opinion. Traditional detection methods based on rules and manual analysis are limited and ineffective against advanced bots.

To address this, the project proposes an AI-based system using Graph Neural Networks (GNNs) and contrastive learning to detect bots more accurately. The system models social networks as graphs, where users are nodes and their interactions are edges, allowing it to analyze both user behavior and relationships.

The methodology involves collecting and preprocessing social media data, constructing interaction graphs, and generating multiple graph views through augmentation. The model learns meaningful user patterns and uses these to classify accounts as bots or genuine users. Performance is evaluated using metrics like accuracy and F1-score.

Conclusion

In conclusion, this research successfully developed an automated social bot detection system using graph-based machine learning techniques. The proposed framework integrates data preprocessing, graph construction, contrastive learning, and classification into a unified pipeline capable of detecting automated accounts in social media networks.By leveraging graph representations and contrastive learning strategies, the model captures both structural and behavioral patterns of user interactions, leading to improved detection accuracy. The system demonstrates strong performance when evaluated using standard classification metrics such as accuracy, precision, recall, and F1-score. The developed framework provides a practical tool for researchers and social media administrators to monitor suspicious activities and maintain the integrity of online platforms. Future work may focus on extending the system to support real-time monitoring and cross-platform analysis for enhanced social media security.

References

[1] W. E. Zhang, Q. Z. Sheng, A. Alhazmi, and C. Li, ‘‘Adversarial attacks on deep-learning models in natural language processing: A survey,’’ ACM Trans. Intell. Syst. Technol. (TIST), vol. 11, no. 3, pp. 1–41, 2020. [2] D. I. Adelani, H. Mai, F. Fang, H. H. Nguyen, J. Yamagishi, and I. Echizen, ‘‘Generating sentiment-preserving fake online reviews using neural language models and their human-and machine-based detection,’’ in Proc. 34th Int. Conf. Adv. Inf. Netw. Appl., 2020, pp. 1341–1354. [3] M. Aljabri, R. Zagrouba, A. Shaahid, F. Alnasser, A. Saleh, and D. M. Alomari, Machine Learning—Based Social Media Bot Detection: A Comprehensive Literature Review. Cham, Switzerland: Springer, 2023. [4] Ali and A. M. Syed, ‘‘Cyberbullying detection using machine learning,’’ Pakistan J. Eng. Technol., vol. 3, no. 2, pp. 45–50, Apr. 2022. [5] J.Wise. Twitter Bots Percentage: How Many Bots are on Twitter? Earth Web. Accessed: Jun. 14, 2024. [Online]. Available: https://earthweb.com/blog/how-many-bots-are-on-twitter/ [6] W. Yue and L. Li, ‘‘Sentiment analysis using Word2vec-CNN-BiLSTM classification,’’ in Proc. 7th Int. Conf. Social Netw. Anal., Manage. Secur. (SNAMS), Dec. 2020, pp. 1–5. [7] J. Pennington, R. Socher, and C. Manning, ‘‘Glove: Global vectors for word representation,’’ in Proc. Conf. Empirical Methods Natural Lang. Process. (EMNLP), 2014, pp. 1532–1543. [8] S. S. Roy, A. I. Awad, L. A. Amare, M. T. Erkihun, and M. Anas, ‘‘Multimodel phishing URL detection using LSTM, bidirectional LSTM, and GRU models,’’ Future Internet, vol. 14, no. 11, p. 340, Nov. 2022. [9] T. B. Brown et al., ‘‘Language models are few-shot learners,’’ in Proc. NIPS, 2020, pp. 1877–1901. [10] M. Tezgider, B. Yildiz, and G. Aydin, ‘‘Text classification using improved bidirectional transformer,’’ Concurrency Comput., Pract. Exper., vol. 34, no. 9, p. 6486, Apr. 2022. [11] S. Kumar and A. Solanki, ‘‘An abstractive text summarization technique using transformer model with self-attention mechanism,’’ Neural Comput. Appl., vol. 35, no. 25, pp. 18603–18622, Sep. 2023.

Copyright

Copyright © 2026 P. Sujitha, J. Yogasri, V. Vysyaa, S. Venkata Lakshmi. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.