A differential privacy processing and publishing method for social network data

Abstract

The invention discloses a differential privacy processing and publishing method for social network data. The method utilizes the fast community detection and differential noising to identify the structural labels of the social network and generate the node labels which make the community nodes gather. The method is based on the adjacency matrix processing of the social network graph. At the same time, the data-independent adaptive method and binary tree structure are used to determine the dense region of the generated upper triangular adjacency matrix. Finally, the noisy adjacency matrix is reconstructed by matrix processing and the network graph is published. The invention introduces the concept of community grouping, which not only protects the data privacy of the social network but alsoensures the better data functionality, and the method of using the upper triangular matrix to reconstruct the density of the sub-region can effectively improve the data processing efficiency, and theoptimal noise-adding edge distribution method designed according to different densities also ensures the privacy protection degree of the scheme.

Description

technical field [0001] The present invention relates to the technical field of differential privacy processing of social network data publishing, in particular to a social network data-oriented privacy processing publishing method based on community detection density aggregation. Background technique [0002] Nowadays, with the popularization of mobile terminal equipment and the development of related mobile network technology, social network and big data technology are getting deeper and deeper into the daily life of ordinary users, but the social relationship and personal privacy data of users in social network Security is still a major concern. [0003] A social network can exist in the form of a graph structure. There is a central node with a large degree (take the star in the Weibo network as an example), and the central node will lead to many edges (followers follow), resulting in less degree Nodes, there can be multiple central nodes in the social network graph, and ...

AI Technical Summary

Problems solved by technology

Compared with the clustering method, the structure modification method can maintain the original scale of the social network, the data defect is relatively small, and the relatively high data utility can be obtained, but the privacy protection ability is relatively reduced

[0004] At the same time, some existing privacy protection models mainly used to solve relational data publishing, including K-anonymity, L-diversity, t-closness, etc., cannot be directly used to solve the problem of privacy leakage in social networks. At the same time, these The method lacks a strict attack model and cannot resist attacks based on background knowledge

Images