∊-k anonymization and adversarial training of graph neural networks for privacy preservation in social networks

作者：

Highlights：

• A novel graph anonymization method for social networks.

• An adversarial training mechanism of graph neural networks (GNNs) to retain as much task performance as possible on anonymous social network analysis.

• A two-stage method covering the data publication and downstream GNN training.

摘要

•A novel graph anonymization method for social networks.•An adversarial training mechanism of graph neural networks (GNNs) to retain as much task performance as possible on anonymous social network analysis.•A two-stage method covering the data publication and downstream GNN training.

论文关键词：Privacy preservation,Anonymization,Graph neural networks,Social network

论文评审过程：Received 2 November 2020, Revised 22 June 2021, Accepted 25 October 2021, Available online 1 November 2021, Version of Record 11 November 2021.

论文官网地址：https://doi.org/10.1016/j.elerap.2021.101105