Abstract: Recently,anonymity model is one of the hot topic techniques in privacy preserving research．The mainly research is how to avoid leakage of sensitive data in data publishing,but also ensures the efficient use of data．This paper proposed a personalized(α[s],k) - lossy decomposition anonymity model．This method publishes the personalized data though generalization technology and personalizedα restriction for different code of the generalization tree．Based on the idea of lossy decomposition in database,data is projected into the sensitive information table and non sensitive information table,and then the redundant information can be realized privacy protection．Experimental results show that the model can provide better privacy.

Key words: Data dissemination,Privacy preserving,Data mining,Lossy join,k-anonymity