Differential Privacy for Edge Weights in Social Networks

Abstract

Social networks can be analyzed to discover important social issues; however, it will cause privacy disclosure in the process. The edge weights play an important role in social graphs, which are associated with sensitive information (e.g., the price of commercial trade). In the paper, we propose the MB-CI (Merging Barrels and Consistency Inference) strategy to protect weighted social graphs. By viewing the edge-weight sequence as an unattributed histogram, differential privacy for edge weights can be implemented based on the histogram. Considering that some edges have the same weight in a social network, we merge the barrels with the same count into one group to reduce the noise required. Moreover,k-indistinguishability between groups is proposed to fulfill differential privacy not to be violated, because simple merging operation may disclose some information by the magnitude of noise itself. For keeping most of the shortest paths unchanged, we do consistency inference according to original order of the sequence as an important postprocessing step. Experimental results show that the proposed approach effectively improved the accuracy and utility of the released data.

SNI: Supervised Anonymization Technique to Publish Social Networks Having Multiple Sensitive Labels

“…To solve VLC, any conventional anonymization methods can be applied, like k-anonymity [4,18,19], ldiversity [20], and di erential privacy [6] in social networks.…”

“…Clearly, to solve VLC, we can insert some additional edges among vertices [4] or insert additional vertices in the graph [5]. For example, by adding an additional edge (edge [6][7][8], the graph in Figure 2 presents that the vertex 7 (Ada) is not unique. Visibly, this action hides the vertex 7 (Ada) among two vertices 2 and 6.…”

“…In this mode, LLC occurs when the sensitive values on the edge of a target victim are estimated, like the recent example. Clearly, according to Figure 3, to solve LLC on edge, we can insert a few additional edges, like edge 7-9, among some vertices or change the values of a few labels on some edges [5,6]. Visibly, this action hides the sensitive labels on edges of Ada among other edges.…”

“…Note that LLC on edge leads to the disclosure of sensitive labels on edges. To solve LLC on edge, many existing methods can be used, like di erential privacy for edge weights in social networks [6]. (2) LLC on Vertex.…”

“…In 2017, Li et al [6] presented a model named "differential privacy method for edge weights in social networks." In this method, a certain amount of random noise is added to the answer of the query set.…”