Research On Privacy Protection Of Soldier Training Data Based On K-Anonymity

With the rapid development of related technologies such as communication technology,network technology,database technology,and software engineering,and the gradual advancement of the modernization of the armed forces,soldiers' training and management are constantly exploring and improving.At present,a complete soldier has basically formed.Training management system,how to ensure the use of information resources of soldiers,while preventing the potential leakage of private information is the focus of the army's focus is also an important subject for researchers.In recent years,Algorithm based on k-anonymity model have been applied to a large number of researches and applications in the fields of medical systems,data mining,and view publishing,and have achieved certain results.However,research and applications on some government or military systems have also been conducted.a bit less.In order to solve the problem of the disclosure of the soldier's data privacy in the troop's soldier training management system,based on the actual troop training data,combined with the application of the anonymous algorithm in the medical system,the algorithm was appropriately modified to apply to the soldiers' training system.While ensuring the smooth use of information technology,we must also strengthen data privacy protection.The main research contents of this topic include:1.From the perspective of privacy protection for soldier training data release,this thesis analyzes the implementation process of the commonly used k-anonymity model,and proposes an improved method for the k-anonymity clustering problem,and analyzes and improves the improved method.2.For the problem that clustering k-anonymity can not resist the homogeneity attack and over-generalization of(k,l)-diversification model,the clustering-based(k,l)-anonymity model is introduced to give the correlation distance.The definition and information missing formulas are used to cluster the attributes of the original data set,and the tuples of similar attributes are clustered to form an equivalent group.Then the equivalent group is generalized and the sensitive attributes are judged according to the diversification requirements.Satisfying a(k,l)-anonymity data set and giving an algorithm for implementing the model.3.Optimize the screening of classic Adult data sets to generate data sets that conforms soldiers' training.Then the model was tested from both the execution time and the information loss of the algorithm to verify the availability of the model.Then under the same data set,the new model is compared with the(k,l)-diversity model and the improved clustering k-anonymity model in terms of algorithm execution time and information missing to verify the superiority of the new model.The(k,l)-anonymity model based on clustering proposed in this thesis has certain application value in soldier soldier's privacy protection and publication data availability in the soldier training data publication.