Differential privacy release method and system for associated tuple data

The invention provides a differential privacy release method and system for associated tuple data, and relates to the field of data mining and privacy protection, which comprises the following steps of: generating initialization noise and disturbing an initialization query result, generating noise meeting a specific auto-covariance matrix according to a new query function, disturbing the query result, finally, processing all queries by adopting an iterative mechanism until the query function sequence is processed, and outputting and publishing a disturbance query result. According to the generalized Laplace noise generation method and the practical and efficient iteration and updating mechanism provided by the invention, the problem that the privacy protection strength is reduced due to the fact that noise generated by differential privacy is independent when tuple data to be protected is related in an existing publishing method of utilizing differential privacy protection associated tuple data is solved.