Sparsity-resilient QoS prediction via ε-DP enhanced subgraph-inductive GNNs in internet of services

Key Points

• Quality of service prediction benefits from subgraph-inductive GNN approaches, enhancing resilience to data sparsity.
• The enhanced model utilizes ε-differential privacy to secure sensitive information while improving prediction accuracy.
• Application of the model addresses challenges related to quality of service in the expanding internet of services landscape.
• The findings suggest that incorporating differential privacy may advocate for broader use of GNNs in sensitive data areas.