Studying the Impact of Edge Privacy on Link Prediction in Temporal Graphs

Dynamic graphs are essential for analyzing complex systems like social or communication networks, allowing researchers to study behaviors and evolution over time. Edge privacy, a key concern in dynamic graphs, involves safeguarding sensitive information about individual connections while the network structure evolves. This paper explores the feasibility of protecting dynamic graphs with differential privacy and using them for effective link prediction, emphasizing the importance of integrating privacy measures into dynamic graph analysis. We evaluate the performance of link prediction algorithms on protected graphs, demonstrating how privacy-enhancing techniques can bolster the robustness and confidentiality of link prediction within evolving network environments. Our study contributes towards establishing more secure and dependable analyses of dynamic network structures by showcasing the practical benefits of edge privacy in link prediction tasks.