Theoretical Study Of Privacy Preserving Based On Data Perturbation In Complete Space And Its Application

The rapid development of network information technology opens a new era of intelligent information processing. Information privacy protection gradually attracts scholars’ wide attention due to the increasing share of data resources. The research of data perturbation privacy-preserving methods and theories has become an important research area of information security, and applying privacy-preserving methods and theories in social networks is a new research hot spot. However, current theoretical study of data perturbation privacy-preserving algorithm is not perfect, especially related theoretical basis. The research of the theoretical basis will provide a better and more solid framework foundation for development, evaluation and practice of data perturbation privacy-preserving technology. In view of this, this paper studies the theory of data perturbation privacy-preserving algorithm, proposes new data perturbation privacy-preserving methods, builds a new data perturbation model which will be applied in the privacy-preserving social networks.The main contributions of this dissertation can be summarized as follows:1. Study the social network of privacy protection issues, prove the effectiveness of perturbation theory in application of social networking; this dissertation first proposes the multi-edge greedy perturbation algorithm for privacy preserving. The simulation shows the feasibility of the results. The research results have successfully solved the social network of multi-edge disturbance privacy protection issues.2. The connection edge weights and its privacy protection are very important in social network privacy protection. Considering of this, we propose a unilateral greedy perturbation algorithm for privacy protection, and prove the technology that disturbing the weight of each edge while saving the shortest path and the length of each node does not exist.3. The dissertation research data perturbation privacy-preserving methods and theories. It proposes a data perturbation method of privacy protection based on Gaussian random multiplication. It means to reasonably choose various parameters of Gaussian distributions to disturbant edge weights of unilateral undirected acyclic network diagram in order to realize the the privacy preservation in social networks. The feasibility of the algorithm is proved theoretically. Simulation experiment results show that the randomization multiplication perturbing method can achieve the anticipated analysis.4. Based on the fixed point theory of the complete lattice, The dissertation gives a proof of how to find out a closed itemset is a process to establish a Galois contact and how to mine all the closed itemsets is the way to find out all the fixed points of its closed Galois operator; therefore, gives theoretical foundation of privacy protection associated algorithm.