Frequent Tables for Fast K-Anonymization

K-anonymization is an important approach to protect data privacy in data publishing. When there are multiple constraints, K-anonymizing a table to satisfy all the constraints is much more complex. We propose a K-anonymization approach, FTB-Classfly, which is based on frequent tables. In stead of generalizing all values in an attribute, FTB-Classfly only generalizes partial tuples that do not satisfy the constraints. By using frequent tables, FTB-Classfly provides higher efficiency than existing approaches. Experimental results show that the proposed FTB-Classfly approach can generate a published table more efficiently than other approaches