A Data Publishing Approaches Of Resisting The Associated Attack On Multi-sensitive Attributes

With the development of data mining technology,a large amount of data has been shared,published used for data analysis which include lots of personal related information.How to prevent leakage of personal privacy under the premise of guaranteeing the release data validity has been a major concern of researchers.But the existing privacy protection technology for multiple sensitive attributes tends to ignore the correlation information between properties which results in a large number of privacy information leak.On the basis of studying multi-dimensional bucket model,Rating model and some other classic privacy protection technology,a model-(l,m)-multi-attribute anonymous models which can well resist the associated attack on multi-sensitive attributes data set is proposed in this paper to solve the privacy information leak problem because of ignoreing the correlation information between properties.First,the paper will find the associated information between the properties and generates association rule set;Secondly,the paper will generate generalization bucket based on different data sets and use the association rule set to filter the generalization bucket to produce generalized table;Next,the paper will user the generalized table to generalize the data set and get the remaining tuples which can not be generalized.Finally,the paper will generalize the remaining tuples separately,and merges the resulting data to produce the final result data set.This paper will use real data sets to verify the algorithm.Experimental results show that the model presented in this paper not only can well resist associated attacks,but also has smaller loss of the additional information and better occult rate,which indicates that the algorithm has better performance.