BotTrans: A Multi-Source Graph Domain Adaptation Approach for Social Bot Detection

2025-06-12

AI Summary

To address the challenges of scarce target-domain labels, pronounced graph heterophily, and poor robustness in single-source domain adaptation for social bot detection, this paper proposes BotTrans, a multi-source graph domain adaptation framework. Methodologically, BotTrans integrates graph neural networks (GNNs), attention-weighted neighborhood aggregation, and multi-source domain adaptation techniques. Its key contributions are: (1) a novel cross-source homogenization topology construction mechanism to mitigate heterophily-induced interference; (2) a source–target correlation-aware weighted multi-source knowledge transfer strategy; and (3) a target-domain semantic consistency regularization module coupled with result refinement. Evaluated on multiple real-world social network datasets, BotTrans achieves an average 5.2% improvement in F1-score over state-of-the-art methods. Notably, it maintains high performance even under extremely low-resource settings, demonstrating superior generalization and robustness.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Application Domains: Social Networks
• Natural Language Processing: Safety and Robustness

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Transferring extensive knowledge from relevant social networks has emerged as a promising solution to overcome label scarcity in detecting social bots and other anomalies with GNN-based models. However, effective transfer faces two critical challenges. Firstly, the network heterophily problem, which is caused by bots hiding malicious behaviors via indiscriminately interacting with human users, hinders the model's ability to learn sufficient and accurate bot-related knowledge from source domains. Secondly, single-source transfer might lead to inferior and unstable results, as the source network may embody weak relevance to the task and provide limited knowledge. To address these challenges, we explore multiple source domains and propose a multi-source graph domain adaptation model named extit{BotTrans}. We initially leverage the labeling knowledge shared across multiple source networks to establish a cross-source-domain topology with increased network homophily. We then aggregate cross-domain neighbor information to enhance the discriminability of source node embeddings. Subsequently, we integrate the relevance between each source-target pair with model optimization, which facilitates knowledge transfer from source networks that are more relevant to the detection task. Additionally, we propose a refinement strategy to improve detection performance by utilizing semantic knowledge within the target domain. Extensive experiments on real-world datasets demonstrate that extit{BotTrans} outperforms the existing state-of-the-art methods, revealing its efficacy in leveraging multi-source knowledge when the target detection task is unlabeled.

Problem

• Overcoming label scarcity in social bot detection with GNNs
• Addressing network heterophily from bots hiding malicious behaviors
• Improving transfer stability via multi-source domain adaptation

Innovation

• Multi-source graph domain adaptation model
• Cross-source-domain topology for homophily
• Relevance-based source-target knowledge transfer