Privacy preserving research for re-publication multiple sensitive attributes in data

Previous works about privacy preserving data publication have most focused on static dataset, which have no update and need “one-time” releases. Only a little of literature has considered the serial data publication on dynamic dataset, but none of them consider perfectly. They can not against various kind of background, or the utility for serial data publishing is low. Based on theoretical analysis, we develop a new generalization principle that effectively limits the risk of Multiple Sensitive Attributes privacy disclosure in re-publication. The results show that our algorithm has higher degree of privacy protection and lower hiding rate.