首页    期刊浏览 2024年11月23日 星期六
登录注册

文章基本信息

  • 标题:Preserving Differential Privacy in Degree-Correlation based Graph Generation
  • 本地全文:下载
  • 作者:Yue Wang ; Xintao Wu
  • 期刊名称:Transactions on Data Privacy
  • 印刷版ISSN:1888-5063
  • 电子版ISSN:2013-1631
  • 出版年度:2013
  • 卷号:6
  • 期号:2
  • 页码:127-145
  • 出版社:IIIA-CSIC
  • 摘要:

    Enabling accurate analysis of social network data while preserving differential privacy has been challenging since graph features such as cluster coefficient often have high sensitivity, which is different from traditional aggregate functions (e.g., count and sum) on tabular data. In this paper, we study the problem of enforcing edge differential privacy in graph generation. The idea is to enforce differential privacy on graph model parameters learned from the original network and then generate the graphs for releasing using the graph model with the private parameters. In particular, we develop a differential privacy preserving graph generator based on the dK-graph generation model. We first derive from the original graph various parameters (i.e., degree correlations) used in the dK-graph model, then enforce edge differential privacy on the learned parameters, and finally use the dKgraph model with the perturbed parameters to generate graphs. For the 2K-graph model, we enforce the edge differential privacy by calibrating noise based on the smooth sensitivity, rather than the global sensitivity. By doing this, we achieve the strict differential privacy guarantee with smaller magnitude noise. We conduct experiments on four real networks and compare the performance of our private dK-graph models with the stochastic Kronecker graph generation model in terms of utility and privacy tradeoff. Empirical evaluations show the developed private dK-graph generation models significantly outperform the approach based on the stochastic Kronecker generation model.

国家哲学社会科学文献中心版权所有