A Study Of Privacy-preserving In Data Publishing Based On Anonymity Model

With the rapid development of computer technology and net technology, a large amount of personal information, involving much privacy of the individual concerned, is stored and released by the government, commercial organizations, etc. However, sharing of the information brings about enormous threat as well as considerable benefits to people. Therefore, how to protect the privacy of the personal information while data being released has become the heated research topic these years.The thesis to begin with, introduces the background, the significance and the current state of the present research. Next, based on the review of the related model and algorithm of the privacy protection in the data release, the thesis proposes an (αi,k)-anonymity model based on sensitivity level aiming at the relational data in table form and a k- anonymity model based on Vertex 1-neighborhood subgraph serving the social net after the study of data privacy protection technology released in the forms of relational data in table and social net respectively. The thesis coves the three main aspects as follows:1. The thesis analyzes the related studies on privacy protection aiming at the data release, gives an overview of the relevant knowledge of privacy protection, and contrasts a series of models and algorithm of the privacy protection in the two forms of data.2. As for the defects of (a,k)-anonymity model, the thesis puts forward an (αi,k)-anonymity model. Describing the sensitivity difference among the sensitivity value by defining its distance, the researcher designs a model employing the strategy of (αi,k)-anonymity clustering algorithm, with the improvement of the conventional (α,k)-anonymity model and the introduction of a concept of undermined connection. The experimental result indicates that the new model, enhancing the performance efficiency, decreasing the loss of information greatly, and better resisting the homogeneity attacks and partial background knowledge attacks, is a more effective privacy protection approach.3. As for the defects of current models in publishing of social network data, the thesis advances a k-anonymity model based on Vertex 1-neighborhood subgraph. Integrating the k-anonymity protection concept into the neighborhood subgraph of social net vertex, the researcher designs a model employing the social net k-anonymity algorithm base on Vertex 1-neighborhood subgraph, which resists the neighborhood background knowledge attacks with the isomorphs among the subgraphs realized by neighborhood anonymization. The experimental result shows that the new model possesses a stronger anti-attack capability.