Sem2Edge: Semantic–Structural Temporal Modeling for Dynamic Link Prediction
PhD. Seminar by: Nahid Abdolrahmanpour Holagh
Date: Thursday, August 20, 2026
Time:11 a.m.
Location: Teams
Abstract:
Dynamic Link Prediction aims to predict future interactions in evolving networks. While recent temporal graph learning methods effectively model structural evolution, semantic information from text is often treated as a static node feature, limiting the ability to capture how meaning changes over time. We propose Sem2Edge, a semantic-aware temporal framework that jointly models semantic evolution and structural dynamics, treating semantic representations as evolving temporal signals that continuously interact with graph structure across snapshots. Evaluated on two real-world dynamic graph datasets (Reddit and Enron), Sem2Edge consistently outperforms structure-only and static-semantic baselines on both classification and ranking metrics, highlighting the importance of temporal semantic modeling for evolving networks.
Keywords: [Insert 3-5 keywords]:
Dynamic Link Prediction, Semantic Evolution, Graph Transformers
PhD Doctoral Committee:
Internal Reader: Dr. Dan Wu
Internal Reader: Dr. Jianguo Lu
External Reader: Dr. Narayan Kar
Advisor (s): Dr. Ziad Kobti
