Research On K-Anonymity For Privacy Preserving

K-Anonymity technique is an important way to protect private information in information sharing. Shared data processed by K-Anonymity can prevent from being attacked by relative linking effectively, avoiding private information leaking.At first, in this paper, according to different privacy preserving objects in information sharing, meta data oriented privacy preserving and data mining knowledge oriented privacy preserving are summarized and analyzed respectively. And the concerning research background about K-Anonymity privacy preserving model is introduced, as well as the implementation technique, representing the K-Anonymity model by instances.Secondly, study and analyze a series of findings of former researchers. There are two key implementation algorithms of K-Anonymity technique, namely full-domain generalization algorithm and local-domain generalization algorithm. Comparative analysis of performance is made between some classic algorithms, and analyzing the advantages and pitfalls of the existing K-Anonymity algorithm. Then study the possible attacks that the K-Anonymity privacy preserving model may undergo theoretically. In this paper five attack methods aimed at K-Anonymity privacy preserving model presenting by instances, are put forward , and then a collection of solutions are proposed, providing the potent safeguard that prevents K-Anonymity privacy preserving model from being attacked.After that, we introduce the main implementation methods' suppression and generalization of K-Anonymity privacy preserving model, and study the precision differences of anonymized data resulting from different generalization methods. The concept ofentropy is brought into K-Anonymity privacy preservingmodel, and a K-Anonymity privacy preserving algorithm, Entropy algorithm, based on entropy classification is put forward. In this paper Entropy algorithm performance is analyzed through experiments, including time complexity analysis and precision analysis of anonymized shared data. The experiment illustrates that Entropy algorithm has been improved more effectively than the former ones.At last, under the some experiment condition, performance tests of Entropy algorithm and the former full-domain generalization algorithm-Basic Incognito algorithm are both carried out. According to experiment results, comparison analysis is made between Entropy algorithm and Basic Incognito algorithm, including time complexity analysis and precision analysis. The analysis outcomes show that although Entropy algorithm needs a longer elapsed time than Basic Incognito, yet Entropy algorithm can provide better anonymized data precision. Entropy algorithm improves the utilization reference value of anonymized data effectively, while assuring the shared data sanctifyingK-Anonymity.