Stability Regularized Graph Embeddings Detect Risk Aware Community Shifts in Online Forums Over Time
Main Article Content
Abstract
The rapid evolution of digital communication has transformed online forums into dynamic ecosystems where user communities constantly form, dissolve, and shift. While these platforms facilitate organic discourse, they also serve as breeding grounds for emergent risks, including coordinated harassment, radicalization, and the dissemination of harmful content. Traditional graph representation learning methods struggle to capture the nuances of temporal community evolution, often producing unstable embeddings that lead to false positives in shift detection. To address this limitation, this paper introduces a novel framework utilizing stability-regularized graph embeddings specifically designed for risk-aware community shift detection. By strictly enforcing temporal consistency in the latent representation space, the proposed methodology distinguishes between natural conversational drift and abrupt structural changes indicative of high-risk behavior. The framework incorporates textual risk signals directly into the structural learning process, enabling a holistic assessment of community dynamics. Extensive analysis demonstrates that applying stability regularization significantly reduces the embedding misalignment commonly observed in sequential graph snapshots. The integration of risk-aware objective constraints further empowers the model to prioritize structural anomalies that correlate with toxic or coordinated malicious activity. Through rigorous evaluation against contemporary baselines, the proposed approach exhibits superior performance in both clustering accuracy and risk identification precision. The findings offer substantial contributions to computational social science and platform moderation, providing a robust mechanism for the early detection of volatile community dynamics in complex social networks.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
All articles published in the Journal of the Artificial Intelligence Engineers Consortium (JAIEC) are licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
This license permits unrestricted use, distribution, reproduction, and adaptation in any medium, provided the original author(s) and source are properly credited.
Authors retain copyright of their work while granting the journal the right of first publication.
References
1. Ma, Y., Qu, D., & Wang, Y. (2026). Tracking evolving communities in fake news cascades using temporal graphs. Scientific Reports.（https://www.nature.com/articles/s41598-026-35175-4）
2. Luo, T., Li, Q., Jiang, M., & Hou, C. (2025). Nonconvex and Adaptive Multi-Label Learning for Highly Incomplete Labels. Knowledge-Based Systems, 114784.
3. Perozzi, B., Al-Rfou, R., & Skiena, S. (2014). DeepWalk: Online learning of social representations. In Proceedings of KDD.
4. Qu, D., Ma, Y., & Pyrozhenko, M. (2026). DISPEL-GNN: De-Illusion via Spectral Stability and Perturbation Bound-Enforced Learning for Community Detection with Risk-Aware Dynamic Attention in Graph Neural Networks. Mathematics, 14(4), 602.（https://doi.org/10.3390/math14040602）
5. Qu, D. Z., & Ma, Y. F. (2025, August). Amon-net: Integrating graph attention and modularity refinement for community detection in complex networks. In 2025 International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA) (pp. 1-5). IEEE.
6. Qu, D., & Ma, Y. (2025, July). Gnc-cut: A hybrid framework for community detection via GNN embeddings and classical clustering. In 2025 6th International Conference on Big Data & Artificial Intelligence & Software Engineering (ICBASE) (pp. 391-395). IEEE.
7. Hamilton, W., Ying, Z., & Leskovec, J. (2017). Inductive representation learning on large graphs. In Advances in Neural Information Processing Systems.
8. Qu, D., & Ma, Y. (2026, March). SC-MGM: soft clustering and Markov-GNN modeling for spatiotemporal community detection in video scene graphs. In Ninth International Conference on Video and Image Processing (ICVIP 2025) (Vol. 14068, pp. 45-55). SPIE.
9. Ma, Y., & Qu, D. (2026, February). TF²-Net: treg-regulated Fourier Transformer for homeostatic community detection in dynamic graphs. In International Conference on Electronic Information Engineering and Artificial Intelligence (EIEAI 2025) (Vol. 14062, pp. 255-260). SPIE.
10. Huang, H., Zhang, J., Zhang, J., Wu, Q., & Xu, C. (2021, May). Ptn: A poisson transfer network for semi-supervised few-shot learning. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 35, No. 2, pp. 1602-1609).
11. Ma, Y., Qu, D., & Wang, Y. (2026). Dynamic community detection using class preserving time series generation with Fourier Markov diffusion. Scientific Reports.（https://doi.org/10.1038/s41598-026-37699-1）
12. Xu, Haoyan, et al. "Glip-ood: Zero-shot graph ood detection with foundation model." arXiv e-prints (2025): arXiv-2504.
13. Qu, D., & Ma, Y. (2025). Magnet-bn: markov-guided Bayesian neural networks for calibrated long-horizon sequence forecasting and community tracking. Mathematics, 13(17), 2740.
14. Xu, H., Yao, Z., Dong, Y., Wang, Z., Rossi, R., Li, M., & Zhao, Y. (2025, September). Few-shot graph out-of-distribution detection with llms. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases (pp. 307-324). Cham: Springer Nature Switzerland.
15. Ma, Y., Qu, D., & Pyrozhenko, M. (2026). Bio-RegNet: A Meta-Homeostatic Bayesian Neural Network Framework Integrating Treg-Inspired Immunoregulation and Autophagic Optimization for Adaptive Community Detection and Stable Intelligence. Biomimetics, 11(1), 48.
16. Xu, Haoyan, et al. "Graph synthetic out-of-distribution exposure with large language models." arXiv preprint arXiv:2504.21198 (2025).
17. Guo, C., Xiao, Z., Liu, Z., & Peng, S. (2020). Rg conditional diagnosability: a novel generalized measure of system-level diagnosis. Theoretical Computer Science, 814, 19-27.
18. Kipf, T. N., & Welling, M. (2017). Semi-supervised classification with graph convolutional networks. In Proceedings of ICLR.
19. Zheng, Z., Meng, L., & Lin, Z. (2025). A Cloud-Based Spatio-Temporal GNN-Transformer Hybrid Model for Traffic Flow Forecasting with External Feature Integration. arXiv preprint arXiv:2510.27039.
20. Xu, Haoyan, et al. "GLIP-OOD: Zero-Shot Graph OOD Detection with Graph Foundation Model." arXiv preprint arXiv:2504.21186 (2025).
21. Qu, D., & Ma, Y. (2026, February). TM-GNN: treg-regulated Markov graph neural networks for stable community detection in dynamic systems. In International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2025) (Vol. 14113, pp. 244-249). SPIE.
22. Qu, D., & Ma, Y. (2026). F2-CommNet: Fourier–Fractional neural networks with Lyapunov stability guarantees for hallucination-resistant community detection. Frontiers in Computational Neuroscience, 19, 1731452. （https://doi.org/10.3389/fncom.2025.1731452）
23. Gu, C., Zhang, W., Huang, Z., Kou, J., Liu, Z., Zhao, C., Liu, C., Zhang, L., Lin, W., Wang, Z., Deng, J., Xie, Y., Huang, G., Zhang, C., Lu, X., Wang, C., Zhang, Z., Yuan, H., Duan, X., & Fang, Y. (2024). LENS: Layers of evaluation of hallucination in GenAI systems. In 2024 7th International Conference on Universal Village (UV) (pp. 1–85). IEEE. https://doi.org/10.1109/UV63228.2024.11189150
24. Yang, Y., Hu, H., Mao, Y., Zhang, J., Wu, C., Jiang, Y., ... & Zhang, C. (2026). OPRIDE: Offline Preference-based Reinforcement Learning via In-Dataset Exploration. arXiv preprint arXiv:2604.02349.
25. Wang, J. (2023, July). A multi-task learning and transfer learning-based model for crop leaf disease identification. In 2023 3rd International Symposium on Computer Technology and Information Science (ISCTIS) (pp. 348-355). IEEE.