[关闭]
@qiutaohanse 2018-07-18T12:04:33.000000Z 字数 2750 阅读 18

去匿名化中的结构相似度

未分类


本文总结了所有已使用的结构相似度,并进行了基本的分类,后期会有相应的实验分析

只考虑无向无权图

首先可以分成两种方式,一种是先构造出特征向量,然后再使用常见的相似度或距离度量来计算,另一种是直接设计一个新的相似度度量方式
分为两大类,第一类是与种子节点无关的,第二类是与种子节点相关的

1.1 基于节点基本信息

1.1.1 节点自身的信息

(1)节点的度[1]
1)使用常见的距离度量来计算距离,然后使用高斯函数
2)RDD(Relative Degree Distance):
(2)中心性[2]
1)closeness centrality
2)betweenness centrality

1.1.2 节点之间的信息

采出部分节点(如k个度最大的节点),然后计算其他每个节点和这些节点的距离
(1)最短路径长度(如Top-K Reference Distance[3]
(2)中心性
1)closeness centrality(如Sampling Closeness Centrality[4])
2)betweenness centrality[5]
(3)eccentricity(衡量唯一性)[6][7]:

,L是其他节点与它的相似度(从大到小排列)

如果部分节点是种子节点:
(1)最短路径长度(如Relative Distance Similarity[8]、Landmark Reference Distance[9]

1.2 基于图局部结构

(1)邻居的度信息
1)1阶度序列,前k个最大的度组成一个向量[10]
2)度分布(直接比较邻居的序列,只适用于噪声极少的情况[11]
(2)k-clique+节点的度+两个节点的共同邻居[12][13][14]
(3)k-邻居度直方图统计[15]

1.3 基于图全局结构

(1)(标准化)拉普拉斯映射[16]

1.4 适用于不同网络上的两个节点的相似度计算(只能直接算,无法先得出特征向量再计算)

(1)度信息+种子节点
1)Inheritance Similarity[17]


2)Dice coefficient[18]:
,传统的,无权重图的权重可以认为是1,在该论文中,给出了一种新的方式:
3)Controversial UMP[19]:

4)Shared Identified Friends[20][21][22]:
5)Subgraph Jacarrd Coefficient[23]:

6)度信息+sif[24]


[1] Structural Data De-Anonymization: Theory and Pratice
[2] Structure based Data De-anonymization of Social Networks and Mobility Traces
[3] Structural Data De-Anonymization: Theory and Pratice
[4] Structural Data De-Anonymization: Theory and Pratice
[5] Deanonymizing Mobility Traces: Using Social Networks as a Side-Channel
[6] De-anonymizing Social Networks
[7] Community-Enhanced De-anonymization of Online Social Networks
[8] Structure based Data De-anonymization of Social Networks and Mobility Traces
[9] Structural Data De-Anonymization: Theory and Pratice
[10] Structural Data De-Anonymization: Theory and Pratice
[11] De-anonymizing D4D Datasets
[12] De-anonymizing Social Networks
[13] An Enhanced Structure-Based De-anonymization of Online Social Networks
[14] Community-Enhanced De-anonymization of Online Social Networks
[15] Blind De-anonymization Attacks using Social Networks
[16] User Identification Across Social Media
[17] Structure based Data De-anonymization of Social Networks and Mobility Traces
[18] Joint Link-Attribute User Identity Resolution in Online Social Networks
[19] Cross-Platform Identification of Anonymous Identical Users in Multiple Social Media Networks
[20] On the Performance of Percolation Graph Matching
[21] An Efficient Reconciliation Algorithm for Social Networks
[22] User Identification Across Social Media
[23] An Enhanced Structure-Based De-anonymization of Online Social Networks
[24] Community-Enhanced De-anonymization of Online Social Networks
添加新批注
在作者公开此批注前,只有你和作者可见。
回复批注