![[2602.14239] A Hybrid TGN-SEAL Model for Dynamic Graph Link Prediction](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.14239] A Hybrid TGN-SEAL Model for Dynamic Graph Link Prediction
Summary
The paper presents a Hybrid TGN-SEAL model aimed at improving link prediction in dynamic graphs, particularly in sparse networks, by integrating structural and temporal information.
Why It Matters
Link prediction in dynamic networks is crucial for various applications, including telecommunications and social networks. This study enhances the predictive accuracy of Temporal Graph Networks (TGNs) by addressing challenges like data sparsity and class imbalance, making it relevant for researchers and practitioners in network science and machine learning.
Key Takeaways
- The Hybrid TGN-SEAL model improves link prediction accuracy by 2.6% over standard TGNs.
- Integrating local topology with temporal information enhances model performance in sparse networks.
- The approach is particularly beneficial for networks with transient interactions, such as telecommunication data.
Computer Science > Social and Information Networks arXiv:2602.14239 (cs) [Submitted on 15 Feb 2026] Title:A Hybrid TGN-SEAL Model for Dynamic Graph Link Prediction Authors:Nafiseh Sadat Sajadi, Behnam Bahrak, Mahdi Jafari Siavoshani View a PDF of the paper titled A Hybrid TGN-SEAL Model for Dynamic Graph Link Prediction, by Nafiseh Sadat Sajadi and 2 other authors View PDF HTML (experimental) Abstract:Predicting links in sparse, continuously evolving networks is a central challenge in network science. Conventional heuristic methods and deep learning models, including Graph Neural Networks (GNNs), are typically designed for static graphs and thus struggle to capture temporal dependencies. Snapshot-based techniques partially address this issue but often encounter data sparsity and class imbalance, particularly in networks with transient interactions such as telecommunication call detail records (CDRs). Temporal Graph Networks (TGNs) model dynamic graphs by updating node embeddings over time; however, their predictive accuracy under sparse conditions remains limited. In this study, we improve the TGN framework by extracting enclosing subgraphs around candidate links, enabling the model to jointly learn structural and temporal information. Experiments on a sparse CDR dataset show that our approach increases average precision by 2.6% over standard TGNs, demonstrating the advantages of integrating local topology for robust link prediction in dynamic networks. Subjects: Social an...
