Xianghua Zeng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xianghua Zeng has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Angsheng Li
Beihang University
3 shared publications - Hao Peng
Beihang University, Beijing, China
3 shared publications - Bin Chong
Peking University, China
1 shared publication - Chunyang Liu
Didi Chuxing, Beijing, China
1 shared publication - Hao Liu
University of Science and Technology of China, Hefei, China
1 shared publication - Philip S. Yu
University of Illinois Chicago
1 shared publication
Latest Publications
- Open Access
- 20 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 20 (9) , 1-32
ROBC TRL : Attacking GNN-Based Social Bot Detectors via Reinforced Manipulation of Bots Control Interaction
Social networks have become a crucial source of real-time information for individuals. The influence of social bots within these platforms has garnered considerable attention from researchers, leading to the development of numerous detection technologies. However, the vulnerability and robustness of these detection methods are still underexplored. Existing Graph Neural Network (GNN) -based methods cannot be directly applied due to the issues of limited control over social agents , the black-box nature of bot detectors , and the heterogeneity of bots . To address these challenges, this article proposes the first adversarial Multi-Agent Reinforcement Learning (MARL) framework for social Bot control attacks ( RoBCtrl ) targeting GNN-based social bot detectors. Specifically, we use a diffusion model to generate high-fidelity bot accounts by reconstructing existing account data with minor modifications, thereby evading detection on social platforms. To the best of our knowledge, this is the first application of diffusion models to mimic the behavior of evolving social bots effectively. We then employ a MARL method to simulate bots’ adversarial behavior. We categorize social accounts based on their influence and budget. Different agents are then employed to control bot accounts across various categories, optimizing the attachment strategy through reinforcement learning. Additionally, a hierarchical state abstraction based on structural entropy is designed to accelerate the reinforcement learning. Extensive experiments on social bot detection datasets demonstrate that our framework can effectively undermine the performance of GNN-based detectors.
- 20 October 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Knowledge Discovery From Data
- Vol. 20 (9) , 1-32
Open AccessROBC TRL : Attacking GNN-Based Social Bot Detectors via Reinforced Manipulation of Bots Control Interaction
Social networks have become a crucial source of real-time information for individuals. The influence of social bots within these platforms has garnered considerable attention from researchers, leading to the development of numerous detection technologies. However, the vulnerability and robustness of these detection methods are still underexplored. Existing Graph Neural Network (GNN) -based methods cannot be directly applied due to the issues of limited control over social agents , the black-box nature of bot detectors , and the heterogeneity of bots . To address these challenges, this article proposes the first adversarial Multi-Agent Reinforcement Learning (MARL) framework for social Bot control attacks ( RoBCtrl ) targeting GNN-based social bot detectors. Specifically, we use a diffusion model to generate high-fidelity bot accounts by reconstructing existing account data with minor modifications, thereby evading detection on social platforms. To the best of our knowledge, this is the first application of diffusion models to mimic the behavior of evolving social bots effectively. We then employ a MARL method to simulate bots’ adversarial behavior. We categorize social accounts based on their influence and budget. Different agents are then employed to control bot accounts across various categories, optimizing the attachment strategy through reinforcement learning. Additionally, a hierarchical state abstraction based on structural entropy is designed to accelerate the reinforcement learning. Extensive experiments on social bot detection datasets demonstrate that our framework can effectively undermine the performance of GNN-based detectors.
- 22 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (4) , 4726-4743
Proactive Bot Detection Based on Structural Information Principles
Bot detection is crucial for combating misinformation and preserving the authenticity of online interactions on social media. However, the increasing sophistication of bots in mimicking genuine accounts and evading detection has created an ongoing arms race between detection systems and modeling techniques. In this paper, we propose a novel Structural Information principles-based Adversarial framework, namely SIAMD, designed to Model bot behaviors and achieve proactive Detection. This framework begins by organizing multi-relational interactions between user accounts and social messages into a unified heterogeneous structure, incorporating structural entropy to quantify the uncertainty inherent in historical activities. The high-dimensional entropy is then minimized to uncover a layered hierarchy within account communities, which facilitates activity determination and account selection in behavioral modeling for bot accounts. For each modeled bot and its selected account, SIAMD extracts historical messages and user descriptions to construct prompts and integrates large language models to generate the associated message content. By embedding synthetic message vertices and establishing multi-relational interactions within the original heterogeneous network, SIAMD achieves network evolution in both structure and content, thereby enhancing graph-based proactive detection in an adversarial manner. Extensive comparative experiments on well-established real-world datasets demonstrate that SIAMD significantly and consistently outperforms state-of-the-art detection baselines for social bots in terms of effectiveness, generalizability, robustness, and interpretability.
- 22 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (4) , 4726-4743
Proactive Bot Detection Based on Structural Information Principles
Bot detection is crucial for combating misinformation and preserving the authenticity of online interactions on social media. However, the increasing sophistication of bots in mimicking genuine accounts and evading detection has created an ongoing arms race between detection systems and modeling techniques. In this paper, we propose a novel Structural Information principles-based Adversarial framework, namely SIAMD, designed to Model bot behaviors and achieve proactive Detection. This framework begins by organizing multi-relational interactions between user accounts and social messages into a unified heterogeneous structure, incorporating structural entropy to quantify the uncertainty inherent in historical activities. The high-dimensional entropy is then minimized to uncover a layered hierarchy within account communities, which facilitates activity determination and account selection in behavioral modeling for bot accounts. For each modeled bot and its selected account, SIAMD extracts historical messages and user descriptions to construct prompts and integrates large language models to generate the associated message content. By embedding synthetic message vertices and establishing multi-relational interactions within the original heterogeneous network, SIAMD achieves network evolution in both structure and content, thereby enhancing graph-based proactive detection in an adversarial manner. Extensive comparative experiments on well-established real-world datasets demonstrate that SIAMD significantly and consistently outperforms state-of-the-art detection baselines for social bots in terms of effectiveness, generalizability, robustness, and interpretability.
- Open Access
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 3716-3727
Robustness Evaluation of Graph-based News Detection Using Network Structural Information
- 3 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 3716-3727
Open AccessRobustness Evaluation of Graph-based News Detection Using Network Structural Information
- 2 July 2024
- Published by Elsevier in Artificial Intelligence
- Vol. 334
Incremental measurement of structural entropy for dynamic graphs
- 2 July 2024
- Published by Elsevier in Artificial Intelligence
- Vol. 334
Incremental measurement of structural entropy for dynamic graphs