Composite Graph Publication Considering Important Data

• 作者: Hongbin Zhao Lijie Li Qilong Han Yuqing Sun
• 来源期刊: 国际计算机前沿大会会议论文集
• 学科: 社会科学
• 关键词: Differential PRIVACY COMPOSITE GRAPH k TRIANGLE COUNT
• 年，卷（期）: 2017,（1）
• 页码范围: 51-53
• 页数: 3页
• 分类号: C5

摘要

Under the premise to protect the privacy of personal information,publishing valuable graph is a challenging issue in privacy research.Appling differential privacy in graph,most of the work focused on graph structure characteristic values,because the basic of differential privacy is data distortion,it’s hard to get valuable composite graph if we add a large number of random noise into the raw data.In this article,we show the key that influence availability is whether the important data keep original value in a composite graph.We analysis the properties of important data of k triangle count,and provide a new method for synthesis graph publication.We show the application of this method in k triangle count,and the experimental results proved the accuracy of the method.