Abstract: Aiming at the problem of privacy leakages caused by similar attacks,this paper proposes(H,p,k)-anonymous model.By classifying sensitive attributes,the number of tuples with different sensitive level in equivalent classes could meet the set threshold H.An anonymous algorithm MAA-SLIE(micro-aggregation algorithm based on sensitive level information entropy)was designed to satisfy the model.Based on the greedy clustering idea,the algorithm ensured the maximum privacy security index of the equivalence class in the clustering process,improved the diversity of sensitive attributes in the equivalence class,and reduced the risk of privacy leakage and information loss.The rationality and effectiveness of the algorithm were verified through experiments.